research article a genetic algorithm-based approach for...
TRANSCRIPT
Research ArticleA Genetic Algorithm-Based Approach for Single-MachineScheduling with Learning Effect and Release Time
Der-Chiang Li1 Peng-Hsiang Hsu12 and Chih-Chieh Chang3
1 Department of Industrial and Information Management National Cheng Kung University 1 University Road Tainan Taiwan2Department of Business Administration Kang-Ning Junior College of Medical Care and Management Taipei Taiwan3 Research Center for Information Technology Innovation Academic Sinica Taipei Taiwan
Correspondence should be addressed to Peng-Hsiang Hsu r38991036mailnckuedutw
Received 24 August 2013 Revised 6 November 2013 Accepted 29 November 2013 Published 4 March 2014
Academic Editor Chin-Chia Wu
Copyright copy 2014 Der-Chiang Li et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
The way to gain knowledge and experience of producing a product in a firm can be seen as new solution for reducing the unitcost in scheduling problems which is known as ldquolearning effectsrdquo In the scheduling of batch processing machines it is sometimesadvantageous to form a nonfull batch while in other situations it is a better strategy to wait for future job arrivals in order toincrease the fullness of the batch However research with learning effect and release times is relatively unexplored Motivated bythis observation we consider a single-machine problem with learning effect and release times where the objective is to minimizethe total completion times We develop a branch-and-bound algorithm and a genetic algorithm-based heuristic for this problemThe performances of the proposed algorithms are evaluated and compared via computational experiments which showed that ourapproach has superior ability in this scenario
1 Introduction
Learning effect has been considered as the most inten-sive phenomenon since it has been proposed in Biskup[1] The basic concept has been set to fix the processingtime of scheduling sequence job from the first job to thelast job which demonstrated that the processing time canbe improved after continuous learning Moreover someresearches also indicated that learning effect can be regardedas a controllable and important component affect process-ing time in scheduling problems (Vickson Nowicki andZdrzałka and Cheng et al [2ndash4]) Although there weremany researches focusing on this phenomenon all of themsometimes assumed that all jobs were allowed to processon machine in time However the release times must beconsidered in many real-world applications in order tomake this assumption valid For example products in asemiconductor wafer fabrication facilities undergo severalhundreds of manufacturing steps such as reentrant processflows sequence-dependent setups diversity of product mixand batch processing With such complexities it wouldbe a great challenge to meet the customersrsquo requirements
such as different priorities ready times and due dates Inthe presence of unequal ready times the application of anonfull batch would sometimes be advantageous In somecases it would be better to wait for new jobs to arrive toincrease the completeness of the batch (Monch et al [5])Following are some researches that focus on schedulingproblem by considering both learning effect and release times(Lee et al Eren Wu and Liu Toksarı Wu et al Ahmadizarand Hosseini Rudek Li and Hsu Kung et al and Yinet al [6ndash16]) In this paper we would like to discuss thesingle-machine total completion time problem with sum ofprocessing time-based learning and release times given thatit is a topic still to be studied and explored
Rinnooy Kan [17] showed that the same problem withoutlearning consideration was NP-hard unless the release timeswere identical Therefore we apply the branch-and-boundalgorithm and genetic algorithm search for the optimalsolution and near-optimal solutions The results show thatthe branch and bound algorithm has solved the instancesless than or equal to 24 jobs Moreover GA also shows goodperformance in computational experiments
Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2014 Article ID 249493 12 pageshttpdxdoiorg1011552014249493
2 Mathematical Problems in Engineering
The rest of the paper is organized as follows Somelearning effect works are described in Section 2 In Section 3the description of notations and the problem formulation aregiven Some dominance properties and two lower boundsare developed to enhance the search efficiency for theoptimal solution followed by descriptions of the geneticalgorithm and the branch-and-bound algorithms are shownin Section 4 The results of a computational experiment aregiven in Section 5 and the conclusions are given in the lastsection
2 Related Works
There were some related researches dealing with schedulingproblem via learning effect In Heizer and Render [18]authors verified that unit costs decrease while a firm gainsmore product knowledge and experience Cheng and Wang[19] introduced the framework of learning effect in a singlemachine More recently Biskup [20] provided reviews ofstate-of-the-art scheduling Wang et al [21] studied the time-dependent learning effect in scheduling recently and cameup with the same learning model as proposed by Kuo andYang [22] where the job processing time is a function oftotal normal processing time of the previously scheduled jobsCheng et al [23] brought up with a new learning model inwhich the actual processing time of a job depends on boththe jobrsquos scheduled position and the processing times of thejobs that are already processed The two models proposedby Biskup [1] and Koulamas and Kyparisis [24] were furthercombined by Wu and Lee Lee and Wu and Yin et al [25ndash27] Also a new experience-based learning effect modelwhich is based on the S-shaped learning curve where jobprocessing times would be dependent on the experience ofthe processor was introduced and analyzed by Janiak andRudek [28] S-J Yang and D-L Yang [29] investigated anew group learning effect model on scheduling problemaiming to minimize the total processing time Yin et al [30]brought a general learning effect model into the field ofscheduling which states that the actual processing time of ajob is a general function of both the total actual processingtimes of the jobs already processed and the jobrsquos scheduledposition They had shown that the problems of minimizingmakespan and the sum of the 119896th power of completion timecould be solved in polynomial time respectively A single-machine scheduling problem with a truncation learningeffect proposed by Wu et al [31] states that job processingtime depends on the processing times of the jobs alreadyprocessed and a controlled parameter They also showed thatpolynomial time can be used to solve some single-machinescheduling problems J-B Wang and M-Z Wang [32] alsocame up with a revised model based on general learningeffect and proved that some single-machine and flowshopscheduling problems can be solved using polynomial timeLu et al [33] applying different learning effect modelssimultaneously studied several single-machine schedulingproblems and concluded that under the proposed modelsthe scheduling problems of minimization of the makespanthe total completion time and the sum of the 119896th power of
job completion times can be solved in polynomial time J-B Wang and J-J Wang [34] studied learning effect modelwhere the actual processing time of a job is not only a non-increasing function of the total weighted normal processingtimes of the jobs already processed but also a nonincreasingfunction of the jobrsquos scheduled position where the weightis a position-dependent weight They also show that theirapproach can solve the problem in polynomial time Li et al[35] investigated several single-machine problems with atruncated sum of processing times based learning effect thatremain polynomially solvable Cheng et al [36] addressedtwo-machine flowshop scheduling with a truncated learningfunction while minimizing the makespan They applied abranch-and-bound and three heuristic algorithms to derivethe optimal and near-optimal solutionsWu [37] studied two-agent scheduling on a single machine involving the learningeffects and deteriorating jobs simultaneouslyThe objective isto minimize the total weighted completion time of the jobs ofthe first agent with the restriction that no tardy job is allowedfor the second agent
3 Notation and Problem Formulation
Before formulating the problem we first introduce somenotations that will be used throughout the paper
119899 the number of jobs119878 1198781015840 119878lowast 119878
1 the sequences of jobs
119869119894 119869119895 the job 119894 and job 119895
119901119894 the normal processing time of job 119894
119901[119894](119878) the normal processing time for the job sched-
uled in the 119894th position in 119878119901(119894) the 119894th job processing time when they are in a
nondecreasing order119901119895119903 the actual processing time of a given job 119895 if it is
scheduled in the 119903th position119903119894 the release time of job 119894
119903[119894](119878) the release time of a job scheduled in the 119894th
position in 119878119903(119894) the 119894th job release time when they are in a non-
decreasing order119886 the learning ratio where 119886 le 0119862119894(119878) 119862
119894(1198781015840) the completion time of job 119894 in 119878 and 119878
1015840119862119894(119878lowast) the completion time of job 119894 in optimal
schedule 119878lowast119862[119894](119878) 119862
[119894](1198781) the completion time of a job sched-
uled in the 119894th position in 119878 and 1198781
TC(119878) TC(1198781015840) the total completion times of sequen-ces 119878 and 119878
1015840120587 1205871015840 120587119888 the subsequences of jobs
The formulation of the problem is described as followsThere are 119899 jobs to be processed on a single machine Themachine can handle one job at a time and machine idle and
Mathematical Problems in Engineering 3
job preemption are not allowed Each job 119895 has a normalprocessing time 119901
119895and a release time 119903
119895 The general job
learning model is 119901119895119903
= 119901119895(1 + sum
119903minus1
119897=1119901[119897])119886 where 119901
119895119903is
the actual processing time of job 119895 scheduled in the 119903thposition and 119886 le 0 is a learning ratio The objective of thisproblem is to find an optimal schedule 119878
lowast that minimizesthe total completion time that is sum119899
119894=1119862119894(119878lowast) le sum
119899
119894=1119862119894(119878)
for any schedule 119878 Using the standard three-field notation ofGraham et al [38] our scheduling problem can be denotedas 1119901
119895119903= 119901119895(1 + sum
119903minus1
119897=1119901[119897])119886sum119862
4 The Branch-and-Bound andGenetic Algorithms
In this paper we will apply the branch-and-bound andgenetic algorithms search for the optimal solution and obtainnear-optimal solution respectively First in order to facilitatethe searching process and improve the branching procedurewe develop some adjacent pairwise interchange propertiesand two lower bounds to use in branch-and-bound algo-rithmThen the procedure of the genetic algorithms is givenat last
41 Dominance Properties Before presenting the adjacentpairwise interchange properties we provide two lemmaswhich will be used in the proofs of the properties in thesequel
Lemma 1 Let119891(119909) = 1+ 119886119909(1 minus 119909)119886minus1
minus (1 + 119909)119886 then119891(119909) ge
0 for 119886 le 0 and 119909 gt 0
Lemma 2 Let 119892(120582) = (120582 minus 1) + (1 + 120582119909)119886minus 120582(1 + 119909)
119886 then119892(120582) ge 0 for 120582 ge 1 119886 le 0 and 119909 gt 0
To fathom the searching tree we develop some dom-inance properties based on a pairwise interchange of twoadjacent jobs 119869
119894and 119869
119895 Let 119878 = (120587 119869
119894 119869119895 1205871015840) and 119878
1015840=
(120587 119869119895 119869119894 1205871015840) be two sequences in which 120587 and 120587
1015840 denotepartial sequences To show that 119878 dominates 1198781015840 it suffices toshow that 119862
119894(119878) + 119862
119895(119878) le 119862
119895(1198781015840) + 119862119894(1198781015840) and 119862
119895(119878) lt 119862
119894(1198781015840)
In addition let 119905 be the completion time of the last job insubsequence 120587 with (119903 minus 1) jobs
Property 1 If 119901119894lt 119901119895and max119903
119894 119903119895 le 119905 then 119878 dominates
1198781015840
Proof Since max119903119894 119903119895 le 119905 we have
119862119894(119878) = 119905 + 119901
119894(1 +
119903minus1
sum
119897=1
119901[119897])
119886
119862119895(119878) = 119905 + 119901
119894(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119895(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119894)
119886
119862119895(1198781015840) = 119905 + 119901
119895(1 +
119903minus1
sum
119897=1
119901[119897])
119886
119862119894(1198781015840) = 119905 + 119901
119895(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119894(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119895)
119886
(1)
After taking the difference of (1) we have
119862119894(1198781015840) minus 119862119895(119878) = (119901
119895minus 119901119894)(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119894(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119895)
119886
minus 119901119895(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119894)
119886
(2)
On substituting120582 = 119901119895119901119894 119906 = (1+sum
119903minus1
119897=1119901[119897]) and119909 = (119901
119894(1+
sum119899
119897=1119901119897)) into (2) and simplifying it we obtain
119862119894(1198781015840) minus 119862119895(119878) = 119901
119894119906119886[(120582 minus 1) + (1 + 120582119909)
119886minus 120582(1 + 119909)
119886]
(3)
By Lemma 2 with 120582 gt 1 119886 le 0 and 119909 gt 0 we have 119862119894(1198781015840) minus
119862119895(119878) gt 0Moreover after taking the difference of total completion
times (TC) between sequences 119878 and 1198781015840 we have
TC (1198781015840) minus TC (119878)
= [119862119895(1198781015840) + 119862119894(1198781015840)] minus [119862
119894(119878) + 119862
119895(119878)]
= 2 (119901119895minus 119901119894)(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119894(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119895)
119886
minus 119901119895(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119894)
119886
(4)
By (3) it can be easily shown that (4) is nonnegative for 119901119894lt
119901119895 Therefore 119878 dominates 1198781015840
The proofs of Properties 2 to 5 are omitted since they aresimilar to that of Property 1
Property 2 If 119903119894le 119905 le 119903
119895le 119905 + 119901
119894(1 + sum
119903minus1
119897=1119901[119897])119886
and 119901119894lt 119901119895
then 119878 dominates 1198781015840
Property 3 If 119905 ge 119903119894and 119905 + 119901
119894(1 + sum
119903minus1
119897=1119901[119897])119886
lt 119903119895 then 119878
dominates 1198781015840
Property 4 If 119905 le 119903119894le 119903119895 119903119894+ 119901119894(1 + sum
119903minus1
119897=1119901[119897])119886
ge 119903119895 and
119901119894lt 119901119895 then 119878 dominates 1198781015840
Property 5 If 119905 le 119903119894and 119903119894+ 119901119894(1 + sum
119903minus1
119897=1119901[119897])119886
lt 119903119895 then 119878
dominates 1198781015840
4 Mathematical Problems in Engineering
In order to further determine the ordering of the remain-ing unscheduled jobs to further speed up the searchingprocess we provide the following property Assume that 119878 =
(120587 120587119888) is a sequence of jobs where 120587 is the scheduled part
containing 119896 jobs and 120587119888 is the unscheduled part Let 119878
1=
(120587 1205871015840) be the sequence in which the unscheduled jobs are
arranged in a nondecreasing order of job processing timesthat is 119901
(119896+1)le 119901(119896+2)
le sdot sdot sdot le 119901(119899)
Property 6 If 119862[119896](1198781) gt max
119895isin120587119888119903119895 then 119878
1= (120587 120587
1015840)
dominates sequences of the type (120587 120587119888) for any unscheduledsequence 120587119888
Proof Since 119862[119896](1198781) gt max
119895isin120587119888119903119895 it implies that all the
unscheduled jobs are ready to be processed on time 119862[119896](1198781)
To obtain the optimal subsequence let 1198781= (120587 120587
1015840) be the
sequence in which the unscheduled jobs are arranged innondecreasing order of jobs processing times
42 Lower Bounds In this subsection we develop two lowerbounds by using the following lemma from Hardy et al [39]
Lemma 3 Suppose that 119886119894and 119887
119894are two sequences of
numbers The sum sum119899
119894=1119886119894119887119894of products of the corresponding
elements is the least if the sequences are monotonic in theopposite sense
First let 119875119878 be a partial schedule in which the order ofthe first 119896 jobs has been determined and let 119878 be a completeschedule obtained from 119875119878 By definition the completiontime for the (119896 + 1)th job is
119862[119896+1]
(119878) = max 119862[119896]
(119878) 119903[119896+1]
+ 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
ge 119862[119896]
(119878) + 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
(5)
Similarly the completion time for the (119896 + 119895)th job is
119862[119896+119895]
(119878) ge 119862[119896]
(119878) +
119895
sum
119894=1
119901(119896+119894)
(1 +
119896
sum
119897=1
119901[119897]
+
119894minus1
sum
119897=1
119901(119896+119897)
)
119886
1 le 119895 le 119899 minus 119896
(6)
The first term on the right hand side of (6) is known anda lower bound of the total completion time for the partialsequence 119875119878 can be obtained by minimizing the secondterm Since the value of (1 + sum
119896
119897=1119901[119897]
+ sum119894minus1
119897=1119901[119896+119897]
)119886 is a
decreasing function of sum119894minus1119897=1
119901[119896+119897]
the total completion timeis minimized by sequencing the unscheduled jobs accordingto the shortest processing time (SPT) rule according toLemma 3 Consequently the first lower bound is
LB1=
119896
sum
119894=1
119862[119894](119878) +
119899minus119896
sum
119895=1
119862(119895) (7)
where 119862(119895)
= 119862[119896](119878) + sum
119895
119894=1119901(119896+119894)
(1 + sum119896
119897=1119901[119897]
+
sum119894minus1
119897=1119901(119899minus119896+119897minus1)
)119886 On the other hand this lower bound may
not be tight if the release time is long To overcome thissituation a second lower bound is established by takingaccount of the release time The completion time for the(119896 + 1)th job is
119862[119896+1]
(119878) = max 119862[119896]
(119878) 119903[119896+1]
+ 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
ge 119903[119896+1]
(119878) + 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
(8)
Similarly the completion time for the (119896 + 119895)th job is
119862[119896+119895]
(119878) ge 119903[119896+119895]
(119878) + 119901[119896+119895]
(1 +
119896
sum
119897=1
119901[119897]
+
119895minus1
sum
119897=1
119901[119896+119897]
)
119886
1 le 119895 le 119899 minus 119896
(9)
Note that 119862[119896+119895]
(119878) is greater than or equal to 1199031015840
(119896+119895) where
1199031015840
(119896+1)le 1199031015840
(119896+2)le sdot sdot sdot le 119903
1015840
(119899)denote the release times of
the unscheduled jobs arranged in a nondecreasing orderThesecond term on the right hand side of (9) is minimized bythe SPT rule since (1 +sum
119896
119897=1119901[119897]+sum119895minus1
119897=1119901[119896+119897]
)119886 is a decreasing
function of sum119895minus1119897=1
119901[119896+119897]
It follows that we have the followingsecond lower bound
LB2=
119896
sum
119894=1
119862[119894]
+
119899minus119896
sum
119895=1
119862(119895) (10)
where 119862(119895)
= 1199031015840
(119896+119895)(119878) +119901
1015840
(119896+1)(1+sum
119896
119897=1119901[119897]+sum119895minus1
119897=11199011015840
(119899minus119896+119897minus1))119886
Note that 1199011015840(119896+119895)
and 1199031015840
(119896+119895)do not necessarily come from the
same job In order tomake the lower bound tighter we choosethe maximum value from (7) and (10) as the lower bounds of119875119878 That is
LB = max LB1 LB2 (11)
43TheProcedure of Genetic Algorithms Agenetic algorithm(GA) is an optimization method that mimics natural pro-cesses GAwas invented by Holland [40] and themost widelyused to solve numerical optimization problems in a widevariety of application fields including biology economicsengineering business agriculture telecommunications andmanufacturing For example in Goldberg [41] authors usingGA in engineering design problems is reviewed in Gen andCheng [42] Soolaki et al [43] use a GA to solve an airlineboarding problem with linear programming models [44 45]and use genetic algorithms to optimize the parameters for thegiven test collections GAs start evolving by generating aninitial population of chromosomes Then a fitness functionis used to compute the relative fitness of each chromosomeof the population The selection crossover and mutationoperators are used in succession to create a new population
Mathematical Problems in Engineering 5
of chromosomes for the next generation This approach hasgained increasing popularity in solving many combinatorialoptimization problems in a wide variety of different disci-plines
431 Initial Settings In a GA every problem is presented bya code and each code is seen as a geneThe existing genes canbe combined and seen as a chromosome each of which is oneof the feasible solutions to a problem However traditionalrepresentation of GA does not work for scheduling problems(Etiler et al [46]) In dealing with this condition this studyadopts the same method that a structure can describe thejobs as a sequence in the problem To specify our approachseveral initial sequences are adopted In GA
1 jobs are placed
according to the shortest processing times (SPT) first ruleIn GA
2 jobs are arranged in earliest ready times (ERT) first
rule In GA3 jobs are arranged in a nondecreasing order on
the sum of job processing times and ready times Note thatbefore performing GA NEH algorithm (Nawaz et al [47]) isutilized to improve the quality of the solutions obtained fromthe previous rules to reduce many idle periods The processof GA
1 GA2 and GA
3are different initial sequences and use
the same selection crossover mutation operators populationsize and generations to obtain near-optimal solution Inaddition the fourth genetic algorithm denoted as GA
4 is
the best one among GA1 GA2 and GA
3 that is GA
4=
minGA1GA2GA3
In order to avoid rapidly observing a local optimum in asmall population or consume more waiting time in a largeone this study set a suitable population size as 60 (119873 =
60) in a preliminary trial It is also an important work toevaluate the fitness of selected chromosomes that each ofthe chromosomes is included or excluded from a feasiblesolution The main goal of this study is to minimize thetotal completion time Assume that 119878
119894(119905) is the 119894th string
in the 119894th generation and the total completion time of 119878119894(119905)
is sum119899
119895=1119862119895(119878119894(119905)) Then the fitness function of 119878
119894(119905) can be
represented as 119891(119878119894(119905)) Following are the calculations of the
strings in fitness function
119891 (119878119894(119905)) = max
1le119897le119873
119899
sum
119895=1
119862119895(119878119897(119905))
minus
119899
sum
119895=1
119862119895(119878119894(119905))
(12)
Moreover it is also crucial work to ensure that the probabilityof selection for a sequence with lower value of the objectivefunction is higher Thus the probability 119875(119878
119894(119905)) can be
written as follows
119875 (119878119894(119905)) =
119891 (119878119894(119905))
sum119899
119894=1119891 (119878119894(119905))
(13)
432 Operators There are a few operators that are used inthis study Following are the descriptions of those operatorscrossover mutation and selection
(a) CrossoverThis is an operator that exchanges some of thegenes of the selected parents with the main concept being
005 010 015 020 025 030000
001
001
002
002
003
Mea
n er
ror (
)
00040 00192 00108 00103 0020600171Pm
Figure 1 The performance of the genetic algorithms for various 119875119898
at (119899 119886 120579119873 119892) = (20 minus005 05 60 100119899)
that the descendant can inherit the advantages of its parentsThis study applied the linear order crossover operator (LOX)proposed by Falkenauer and Bouffouix [48] and is one of thebetter performers among the others (Etiler et al [46]) Theprobability of crossover is set to 1
(b) Mutation The main object of mutation is to achieve foran overall optimal solution and to avoid a locally optimalone In this study the mutation rates (119875
119898) are set at 010
based on our preliminary experiment as shown in Figure 1For (119899 119886 120579119873 119892) = (20 minus005 05 60 100119899) 100 sets of datawere randomly generated to evaluate the performance of theproposed algorithms with varying values of 119875
119898 The results
showed that the proposed algorithmshad the leastmean errorpercentage at 119875
119898= 010
(c) Selection This is a process that determines the proba-bility of each chromosome and is used to decide the betterchromosomes with the better fitness value The evolutionimplemented in our algorithm is based on the elitist list Wecopy the best offspring and use them to generate some ofthe next generation The rest of the offspring are generatedfrom the parent chromosomes by the roulette wheel selectionmethod which can maintain the variety of genes
433 Stopping Criteria In the preliminary experimentthe proposed GAs are terminated after 100 lowast 119899 genera-tions as shown in Figures 2 and 3 For (119899 119886 120579119873 119875
119898) =
(20 minus005 05 60 010) the above 100 sets of randomly gen-erated data were used to evaluate the performance of theproposed algorithms with varying values of 119892 The resultsshowed that the least mean error percentage of the proposedalgorithms would stabilize with reasonable CPU time rangeafter 119892 = 100119899
5 Computational Experiment
A computational experiment was conducted to evaluatethe efficiency of the branch-and-bound algorithm and
6 Mathematical Problems in Engineering
00738 00510 00315 00258 00040
000
001
002
003
004
005
006
007
008
Mea
n er
ror (
)
40n 60n 80n 100n20n
g
Figure 2 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
01666 03260 04981 06800 08382
000
010
020
030
040
050
060
070
080
090
CPU
tim
es (s
)
g
40n 60n 80n 100n20n
Figure 3 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
the accuracies of the genetic algorithmsThe algorithms werecoded in Fortran and run on Compaq Visual Fortran version66 on an Intel(R)Core(TM)2QuadCPU266GHzwith 4GBRAM on Windows Vista The experimental design followedReeves [49] design The job processing times were generatedfrom a uniform distribution over the integers between 1 and20 in every case while the release times were generated froma uniform distribution over the integers on (0 20119899120579) where 119899is the number of jobs Five different sets of problem instanceswere generated by giving 120579 the values 1119899 025 05 075and 1
For the branch-and-bound algorithm the average and themaximum numbers of nodes as well as the average and themaximum execution times (in seconds) were recorded Forthe three genetic algorithms the mean and the maximum
0
200
400
600
800
1000
1200
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005
minus010
minus015
minus020
120579
1n
Figure 4Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 12
error percentages were recorded where the error percentagewas calculated as
(GA119894minus TClowast)
TClowastlowast 100 (14)
where GA119894(119894 = 1 2 3 4) is the total completion time
obtained from the genetic algorithmand TClowast is the totalcompletion time of the optimal schedule The computationaltimes of the heuristic algorithmswere not recorded since theywere finished within a second
In the computational experiment four different numbersof jobs (119899 = 12 16 20 and 24) four different values of learn-ing effect (119886 = minus005 minus010 minus015 andminus020) and five differ-ent values of generation parameter of release times (120579 = 1119899025 05 075 and 1) were tested in the branch-and-boundalgorithmAs a consequence 80 experimental situationswereexamined A set of 20 instances were randomly generatedfor each situation and a total of 1600 problems were testedThe algorithms were set to skip to the next set of data if thenumber of nodes exceeded 108 The results are presented inTable 1 and Figures 4 5 6 and 7 Figures 4ndash7 showed theaverage number of nodes for various 120579 and 119886 at job size 1216 20 and 24 respectively The average number of nodesdecreased as the value of 120579 increased when 119899 was greaterthan 16 This was the direct result of the efficiency of LB
1
and LB2 As 120579 increased the frequency of applications of LB
2
would increase Consequently it would yield longer releasetimes in those cases and the properties were more powerfulMoreover LB
1is more efficient than LB
2 Table 1 and Figures
4ndash7 also showed whether in job size the algorithms had theleast mean number of nodes at 120579 = 1119899 It was due to thefact that with 120579 = 1119899 the release time was relatively shortand the completion time would readily exceed the release
Mathematical Problems in Engineering 7
0
10000
20000
30000
40000
50000
60000
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005minus010
minus015
minus020
120579
1n
Figure 5The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 16
time In those cases Property 6 was applied more frequentlyconversely the completion time would not easily exceed therelease time when the values of 120579 increased Moreover thenumber of nodes increased exponentially as the number ofjobs increased which was typical of an NP-hard problem Asillustrated in Table 1 when 119899 = 24 there were five cases inwhich the branch-and-bound algorithm could solve all theproblems optimally larger than 10
8 nodes The branch-and-bound algorithm had the worst performance when (119899 119886 120579) =
(24 minus005 025)with 87times107 nodes and 5234 secondsWith
120579 fixed at 1119899 the decrease of the completion time would berelatively small at the beginning when the learning effectwassmall (eg 119886 = minus005) In other words the completion timewould easily exceed the release time which would expeditethe timing of invoking Property 6 and consequently theaverage number of nodes would be smaller With 120579 = 025 as119899 increased the corresponding least average number of nodeswould occur at greater values of learning effect
The performance of the proposed GA algorithms out ofthe 80 evaluations and a total of 1600 problems was testedThe number of times that each of the objective functions ofthe GA
1 GA2 and GA
3algorithms had the smallest mean
error percentage was 45 41 and 49 respectively In additionin Table 1 and Figures 8 9 and 10 their performanceswere not affected with the learning rate the generationparameter 120579 of release times or the number of jobs Noneof the three genetic algorithms had absolutely dominantperformance in terms of mean error percentage Howeverthe combined algorithm GA
4strikingly outperformed each
of the three algorithms in terms of the maximummean error
0
500000
1000000
1500000
2000000
2500000
3000000
3500000
4000000
4500000
5000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 6Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 20
0
5000000
10000000
15000000
20000000
25000000
30000000
35000000
40000000
45000000
50000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 7The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 24
8 Mathematical Problems in Engineering
Table1Th
eperform
ance
oftheb
ranch-
and-bo
undandgenetic
algorithm
s
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
12
minus005
111989942
385
00031
00156
2000858
17151
000
00000
00004
0108013
000
00000
00025
549
1781
00187
00624
2001055
21091
00142
02839
000
00000
00000
00000
00050
473
1189
00140
00312
20000
00000
00000
00000
00000
00000
00000
00000
00075
414
1013
00125
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0569
2427
00164
00624
20000
00000
00000
00000
00000
00000
00000
00000
00
minus010
111989962
488
00023
00156
2001646
20568
01617
20568
000
00000
00000
00000
00025
1047
4318
00312
01092
20000
4400878
000
4400878
00283
05665
000
00000
00050
480
1321
00140
004
6820
000
00000
00006
8413
676
00019
00373
000
00000
00075
786
5291
00203
01404
20000
00000
00000
00000
00000
00000
00000
00000
0010
0396
1382
00125
004
6820
000
00000
00000
00000
00000
00000
00000
00000
00
minus015
111989921
7100016
00156
2000206
04130
00000
00000
00360
04130
00000
00000
025
817
4363
00226
00936
20000
00000
00000
00000
0001228
24567
000
00000
00050
493
1969
00148
00624
20000
00000
00000
00000
00000
00000
00000
00000
00075
286
803
000
9400312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0417
1259
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
111989947
336
00023
00156
2001295
25896
01295
25896
000
03000
69000
00000
00025
1091
6689
00289
01560
2000106
02119
000
00000
00000
00000
00000
00000
00050
337
948
00109
00312
2000314
06284
000
00000
00000
00000
00000
00000
00075
452
1121
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0319
1387
00086
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
16
minus005
111989974
365
00078
00312
2002066
34518
02202
27743
01114
12163
006
0812
163
025
39325
105847
20108
58188
2004522
53071
01479
17548
00621
04629
00309
03126
050
18779
107488
08697
44616
2000052
01035
00935
16983
00849
16983
000
00000
00075
22967
290574
09259
102025
20000
00000
00000
00000
00000
00000
00000
00000
0010
05144
34260
02535
15288
20000
4100811
00115
02306
000
00000
00000
00000
00
minus010
111989991
323
00086
00312
2001813
34958
02368
27957
01906
19030
000
00000
00025
53237
291654
25826
127297
2003151
25393
02824
21606
01918
21606
01271
21606
050
15022
193968
06778
81277
2000616
10305
00154
03088
00154
03088
00000
00000
075
15144
197196
06474
80029
20000
00000
00000
00000
00000
0600120
000
00000
0010
03403
9875
01794
04992
20000
00000
00000
00000
00000
00000
00000
00000
00
minus015
1119899118
1368
00101
01092
2002462
45183
00431
06619
00103
02068
000
00000
00025
52185
4846
6222994
190165
2000762
08319
03212
37507
00969
10191
00418
08319
050
17808
101503
08221
46020
2000740
07668
00356
06773
00574
06773
00339
06773
075
3763
16453
01997
08112
20000
00000
00000
00000
00000
6701343
000
00000
0010
03882
14411
01911
07020
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899200
828
00172
00624
2001832
26781
03397
1766
003644
22420
000
00000
00025
25881
149925
11224
53664
2000140
02772
00299
04614
000
9801922
000
0200034
050
11829
48284
05819
23556
20000
00000
0000072
01431
000
00000
00000
00000
00075
2687
12382
01427
05928
20000
00000
00000
00000
00000
00000
00000
00000
0010
02047
5258
01076
02340
20000
00000
00000
00000
00000
00000
00000
00000
00
Mathematical Problems in Engineering 9
Table1Con
tinued
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
20
minus005
111989982
278
00156
00624
2016
268
74028
03538
69562
00113
02265
000
00000
00025
1319111
4407360
1059941
3224697
2007803
29510
05928
32815
08700
42205
02480
15709
050
963361
9983475
618895
5687017
2000377
040
1601608
1746
402224
15708
00033
00510
075
323776
2518001
229485
1831606
2000022
004
46000
00000
0000027
004
46000
00000
0010
041188
138955
33642
1110
7220
00017
00338
00017
00338
000
00000
00000
00000
00
minus010
11198991345
14123
01716
17317
2006707
34958
04690
28787
04353
34958
00000
00000
025
2134707
16978059
1543318
12044214
2003845
23175
01848
06798
03536
21979
004
6002909
050
202434
1320363
156266
1024927
2000000
00000
000
4500786
00000
00000
00000
00000
075
67733
450297
55645
3452
2920
000
00000
00000
00000
00000
00000
00000
00000
0010
030928
238389
26192
185483
2000023
004
62000
00000
0000023
004
62000
00000
00
minus015
1119899293
2224
004
2902964
2001670
13088
01845
29345
01835
29345
000
00000
00025
1539138
7544
291
1039052
4970
186
20046
8529738
01876
10582
02690
19134
00544
09353
050
235214
2737579
16906
61920840
20000
4600923
00032
006
4100073
01284
000
00000
00075
56901
281822
45716
225576
20000
00000
00000
00000
00000
00000
00000
00000
0010
017992
51529
15163
36973
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899698
4825
00991
06240
2002436
15781
04239
404
2602224
15781
01927
15781
025
434800
664
668017
2750735
40004590
2001596
14118
00549
02968
02765
21869
00203
02249
050
107201
675811
80878
476895
20000
00000
00000
00000
0000058
01155
000
00000
00075
31529
163854
25990
110918
20000
00000
00000
00000
00000
00000
00000
00000
0010
015118
63845
13198
49453
20000
00000
0000016
00324
00016
00318
000
00000
00
24
minus005
1119899478
2082
01248
04836
2004707
39003
02017
20990
01368
18118
00959
18118
025
43205517
875044
4452338760
1060
43535
1304991
21418
06032
19032
046
0720265
02283
13231
050
11780220
66191783
13054869
65948027
1900758
10456
00313
02363
01334
10456
00161
02363
075
1216867
4261994
1562529
4708125
2000056
01126
00198
03359
00174
01871
000
00000
0010
0888475
5439067
1144057
69244
1720
000
00000
00000
00000
00000
00000
00000
00000
00
minus010
11198991074
12050
02496
26364
2007378
42545
03868
24023
02577
18340
00541
06057
025
24953166
5300
6630
28547095
6344
0156
1104324
22382
05843
29043
06883
25478
03452
22382
050
3116322
31612798
3661781
38159717
2000137
0118
400087
00883
00075
01028
000
00000
00075
513685
2435858
672926
2964331
20000
00000
0000015
00306
00112
01371
000
00000
0010
0270431
1462391
35260
91960620
20000
00000
0000052
01037
00039
00771
000
00000
00
minus015
1119899674
2613
01693
06084
2007280
78202
03115
27017
02512
34255
00284
05683
025
23417410
61951876
253876
736371360
415
09250
33429
02948
12902
02232
17855
00499
04354
050
4118014
36622393
4749373
39276216
2000114
0119
300053
01058
000
00000
00000
00000
00075
661765
5958864
874869
7920327
2000073
01458
00075
01509
000
4300867
000
00000
0010
0837932
11995859
968626
13244172
2000000
00000
00000
00000
00000
00000
00000
00000
minus020
11198993126
15554
07082
31980
2001942
32396
04290
33146
03196
32396
01669
32396
025
9333827
72575949
9993
219
80682031
1500589
02822
006
4607914
00844
07331
00204
02057
050
384242
2271989
508384
2808486
2000038
00760
00130
02596
00130
02596
00000
00000
075
570054
4053320
7017
325311990
20000
00000
00000
00000
00000
00000
00000
00000
0010
0233727
1202819
2915
7413260
0620
000
00000
00000
00000
00000
00000
00000
00000
00NoteldquoO
Frdquodeno
testhe
numbero
finstances
in20
setsof
datathatcanbe
solved
inlessthan
108no
desb
yusingtheb
ranch-
and-bo
undmetho
d
10 Mathematical Problems in Engineering
000
005
010
015
020
025
030
GA1 GA2 GA3 GA4GA
n16
n20n12
n24
Mea
n er
ror (
)
Figure 8 The performance of the genetic algorithms for various 119899
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
Mea
n er
ror (
)
minus005
minus010 minus020
minus015
Figure 9 The performance of the genetic algorithms for various 119886
percentage among all the 80 scenarios with varying learningrates numbers of jobs and values of 120579 The maximummean error percentages of GA
1 GA2 GA3 and GA
4were
16268 06032 08700 and 03452 respectively Themaximum mean error percentage of GA
1was more than
four times that of GA4 The combined algorithm GA
4also
clearly outperformed each of the three algorithms in terms
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
02505
0751
1n
Mea
n er
ror (
)
Figure 10The performance of the genetic algorithms for various 120579
of the maximum error percentage The worst cases of thecorresponding algorithms were 78202 69562 42205and 32396 respectively The worst case error of GA
1was
more than twice that of GA4Thus we would recommend the
combined algorithm GA4
6 Conclusions
In this paper we address a single-machine total completiontime problem with the sum of processing times basedlearning effect and release timesThe objective is to minimizethe total completion time The problem without learningconsideration is NP-hard one Thus we develop a branch-and-bound algorithm incorporatingwith several dominancesand two lower bounds to derive optimal solution and geneticalgorithms that was proposed to obtain near-optimal solu-tions respectively
The branch-and-bound algorithm performs well to solvethe instances of less than or equal to 24 jobs in a reasonableamount of time In the different varying parameter theproposed genetic algorithms are quite accurate for small sizeproblems and the mean error percentage of the proposedgenetic algorithms are less than 035 Another interestingtopic for future study is to consider and study the problem inthe multimachine environments or multicriteria cases
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
2 Mathematical Problems in Engineering
The rest of the paper is organized as follows Somelearning effect works are described in Section 2 In Section 3the description of notations and the problem formulation aregiven Some dominance properties and two lower boundsare developed to enhance the search efficiency for theoptimal solution followed by descriptions of the geneticalgorithm and the branch-and-bound algorithms are shownin Section 4 The results of a computational experiment aregiven in Section 5 and the conclusions are given in the lastsection
2 Related Works
There were some related researches dealing with schedulingproblem via learning effect In Heizer and Render [18]authors verified that unit costs decrease while a firm gainsmore product knowledge and experience Cheng and Wang[19] introduced the framework of learning effect in a singlemachine More recently Biskup [20] provided reviews ofstate-of-the-art scheduling Wang et al [21] studied the time-dependent learning effect in scheduling recently and cameup with the same learning model as proposed by Kuo andYang [22] where the job processing time is a function oftotal normal processing time of the previously scheduled jobsCheng et al [23] brought up with a new learning model inwhich the actual processing time of a job depends on boththe jobrsquos scheduled position and the processing times of thejobs that are already processed The two models proposedby Biskup [1] and Koulamas and Kyparisis [24] were furthercombined by Wu and Lee Lee and Wu and Yin et al [25ndash27] Also a new experience-based learning effect modelwhich is based on the S-shaped learning curve where jobprocessing times would be dependent on the experience ofthe processor was introduced and analyzed by Janiak andRudek [28] S-J Yang and D-L Yang [29] investigated anew group learning effect model on scheduling problemaiming to minimize the total processing time Yin et al [30]brought a general learning effect model into the field ofscheduling which states that the actual processing time of ajob is a general function of both the total actual processingtimes of the jobs already processed and the jobrsquos scheduledposition They had shown that the problems of minimizingmakespan and the sum of the 119896th power of completion timecould be solved in polynomial time respectively A single-machine scheduling problem with a truncation learningeffect proposed by Wu et al [31] states that job processingtime depends on the processing times of the jobs alreadyprocessed and a controlled parameter They also showed thatpolynomial time can be used to solve some single-machinescheduling problems J-B Wang and M-Z Wang [32] alsocame up with a revised model based on general learningeffect and proved that some single-machine and flowshopscheduling problems can be solved using polynomial timeLu et al [33] applying different learning effect modelssimultaneously studied several single-machine schedulingproblems and concluded that under the proposed modelsthe scheduling problems of minimization of the makespanthe total completion time and the sum of the 119896th power of
job completion times can be solved in polynomial time J-B Wang and J-J Wang [34] studied learning effect modelwhere the actual processing time of a job is not only a non-increasing function of the total weighted normal processingtimes of the jobs already processed but also a nonincreasingfunction of the jobrsquos scheduled position where the weightis a position-dependent weight They also show that theirapproach can solve the problem in polynomial time Li et al[35] investigated several single-machine problems with atruncated sum of processing times based learning effect thatremain polynomially solvable Cheng et al [36] addressedtwo-machine flowshop scheduling with a truncated learningfunction while minimizing the makespan They applied abranch-and-bound and three heuristic algorithms to derivethe optimal and near-optimal solutionsWu [37] studied two-agent scheduling on a single machine involving the learningeffects and deteriorating jobs simultaneouslyThe objective isto minimize the total weighted completion time of the jobs ofthe first agent with the restriction that no tardy job is allowedfor the second agent
3 Notation and Problem Formulation
Before formulating the problem we first introduce somenotations that will be used throughout the paper
119899 the number of jobs119878 1198781015840 119878lowast 119878
1 the sequences of jobs
119869119894 119869119895 the job 119894 and job 119895
119901119894 the normal processing time of job 119894
119901[119894](119878) the normal processing time for the job sched-
uled in the 119894th position in 119878119901(119894) the 119894th job processing time when they are in a
nondecreasing order119901119895119903 the actual processing time of a given job 119895 if it is
scheduled in the 119903th position119903119894 the release time of job 119894
119903[119894](119878) the release time of a job scheduled in the 119894th
position in 119878119903(119894) the 119894th job release time when they are in a non-
decreasing order119886 the learning ratio where 119886 le 0119862119894(119878) 119862
119894(1198781015840) the completion time of job 119894 in 119878 and 119878
1015840119862119894(119878lowast) the completion time of job 119894 in optimal
schedule 119878lowast119862[119894](119878) 119862
[119894](1198781) the completion time of a job sched-
uled in the 119894th position in 119878 and 1198781
TC(119878) TC(1198781015840) the total completion times of sequen-ces 119878 and 119878
1015840120587 1205871015840 120587119888 the subsequences of jobs
The formulation of the problem is described as followsThere are 119899 jobs to be processed on a single machine Themachine can handle one job at a time and machine idle and
Mathematical Problems in Engineering 3
job preemption are not allowed Each job 119895 has a normalprocessing time 119901
119895and a release time 119903
119895 The general job
learning model is 119901119895119903
= 119901119895(1 + sum
119903minus1
119897=1119901[119897])119886 where 119901
119895119903is
the actual processing time of job 119895 scheduled in the 119903thposition and 119886 le 0 is a learning ratio The objective of thisproblem is to find an optimal schedule 119878
lowast that minimizesthe total completion time that is sum119899
119894=1119862119894(119878lowast) le sum
119899
119894=1119862119894(119878)
for any schedule 119878 Using the standard three-field notation ofGraham et al [38] our scheduling problem can be denotedas 1119901
119895119903= 119901119895(1 + sum
119903minus1
119897=1119901[119897])119886sum119862
4 The Branch-and-Bound andGenetic Algorithms
In this paper we will apply the branch-and-bound andgenetic algorithms search for the optimal solution and obtainnear-optimal solution respectively First in order to facilitatethe searching process and improve the branching procedurewe develop some adjacent pairwise interchange propertiesand two lower bounds to use in branch-and-bound algo-rithmThen the procedure of the genetic algorithms is givenat last
41 Dominance Properties Before presenting the adjacentpairwise interchange properties we provide two lemmaswhich will be used in the proofs of the properties in thesequel
Lemma 1 Let119891(119909) = 1+ 119886119909(1 minus 119909)119886minus1
minus (1 + 119909)119886 then119891(119909) ge
0 for 119886 le 0 and 119909 gt 0
Lemma 2 Let 119892(120582) = (120582 minus 1) + (1 + 120582119909)119886minus 120582(1 + 119909)
119886 then119892(120582) ge 0 for 120582 ge 1 119886 le 0 and 119909 gt 0
To fathom the searching tree we develop some dom-inance properties based on a pairwise interchange of twoadjacent jobs 119869
119894and 119869
119895 Let 119878 = (120587 119869
119894 119869119895 1205871015840) and 119878
1015840=
(120587 119869119895 119869119894 1205871015840) be two sequences in which 120587 and 120587
1015840 denotepartial sequences To show that 119878 dominates 1198781015840 it suffices toshow that 119862
119894(119878) + 119862
119895(119878) le 119862
119895(1198781015840) + 119862119894(1198781015840) and 119862
119895(119878) lt 119862
119894(1198781015840)
In addition let 119905 be the completion time of the last job insubsequence 120587 with (119903 minus 1) jobs
Property 1 If 119901119894lt 119901119895and max119903
119894 119903119895 le 119905 then 119878 dominates
1198781015840
Proof Since max119903119894 119903119895 le 119905 we have
119862119894(119878) = 119905 + 119901
119894(1 +
119903minus1
sum
119897=1
119901[119897])
119886
119862119895(119878) = 119905 + 119901
119894(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119895(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119894)
119886
119862119895(1198781015840) = 119905 + 119901
119895(1 +
119903minus1
sum
119897=1
119901[119897])
119886
119862119894(1198781015840) = 119905 + 119901
119895(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119894(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119895)
119886
(1)
After taking the difference of (1) we have
119862119894(1198781015840) minus 119862119895(119878) = (119901
119895minus 119901119894)(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119894(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119895)
119886
minus 119901119895(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119894)
119886
(2)
On substituting120582 = 119901119895119901119894 119906 = (1+sum
119903minus1
119897=1119901[119897]) and119909 = (119901
119894(1+
sum119899
119897=1119901119897)) into (2) and simplifying it we obtain
119862119894(1198781015840) minus 119862119895(119878) = 119901
119894119906119886[(120582 minus 1) + (1 + 120582119909)
119886minus 120582(1 + 119909)
119886]
(3)
By Lemma 2 with 120582 gt 1 119886 le 0 and 119909 gt 0 we have 119862119894(1198781015840) minus
119862119895(119878) gt 0Moreover after taking the difference of total completion
times (TC) between sequences 119878 and 1198781015840 we have
TC (1198781015840) minus TC (119878)
= [119862119895(1198781015840) + 119862119894(1198781015840)] minus [119862
119894(119878) + 119862
119895(119878)]
= 2 (119901119895minus 119901119894)(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119894(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119895)
119886
minus 119901119895(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119894)
119886
(4)
By (3) it can be easily shown that (4) is nonnegative for 119901119894lt
119901119895 Therefore 119878 dominates 1198781015840
The proofs of Properties 2 to 5 are omitted since they aresimilar to that of Property 1
Property 2 If 119903119894le 119905 le 119903
119895le 119905 + 119901
119894(1 + sum
119903minus1
119897=1119901[119897])119886
and 119901119894lt 119901119895
then 119878 dominates 1198781015840
Property 3 If 119905 ge 119903119894and 119905 + 119901
119894(1 + sum
119903minus1
119897=1119901[119897])119886
lt 119903119895 then 119878
dominates 1198781015840
Property 4 If 119905 le 119903119894le 119903119895 119903119894+ 119901119894(1 + sum
119903minus1
119897=1119901[119897])119886
ge 119903119895 and
119901119894lt 119901119895 then 119878 dominates 1198781015840
Property 5 If 119905 le 119903119894and 119903119894+ 119901119894(1 + sum
119903minus1
119897=1119901[119897])119886
lt 119903119895 then 119878
dominates 1198781015840
4 Mathematical Problems in Engineering
In order to further determine the ordering of the remain-ing unscheduled jobs to further speed up the searchingprocess we provide the following property Assume that 119878 =
(120587 120587119888) is a sequence of jobs where 120587 is the scheduled part
containing 119896 jobs and 120587119888 is the unscheduled part Let 119878
1=
(120587 1205871015840) be the sequence in which the unscheduled jobs are
arranged in a nondecreasing order of job processing timesthat is 119901
(119896+1)le 119901(119896+2)
le sdot sdot sdot le 119901(119899)
Property 6 If 119862[119896](1198781) gt max
119895isin120587119888119903119895 then 119878
1= (120587 120587
1015840)
dominates sequences of the type (120587 120587119888) for any unscheduledsequence 120587119888
Proof Since 119862[119896](1198781) gt max
119895isin120587119888119903119895 it implies that all the
unscheduled jobs are ready to be processed on time 119862[119896](1198781)
To obtain the optimal subsequence let 1198781= (120587 120587
1015840) be the
sequence in which the unscheduled jobs are arranged innondecreasing order of jobs processing times
42 Lower Bounds In this subsection we develop two lowerbounds by using the following lemma from Hardy et al [39]
Lemma 3 Suppose that 119886119894and 119887
119894are two sequences of
numbers The sum sum119899
119894=1119886119894119887119894of products of the corresponding
elements is the least if the sequences are monotonic in theopposite sense
First let 119875119878 be a partial schedule in which the order ofthe first 119896 jobs has been determined and let 119878 be a completeschedule obtained from 119875119878 By definition the completiontime for the (119896 + 1)th job is
119862[119896+1]
(119878) = max 119862[119896]
(119878) 119903[119896+1]
+ 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
ge 119862[119896]
(119878) + 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
(5)
Similarly the completion time for the (119896 + 119895)th job is
119862[119896+119895]
(119878) ge 119862[119896]
(119878) +
119895
sum
119894=1
119901(119896+119894)
(1 +
119896
sum
119897=1
119901[119897]
+
119894minus1
sum
119897=1
119901(119896+119897)
)
119886
1 le 119895 le 119899 minus 119896
(6)
The first term on the right hand side of (6) is known anda lower bound of the total completion time for the partialsequence 119875119878 can be obtained by minimizing the secondterm Since the value of (1 + sum
119896
119897=1119901[119897]
+ sum119894minus1
119897=1119901[119896+119897]
)119886 is a
decreasing function of sum119894minus1119897=1
119901[119896+119897]
the total completion timeis minimized by sequencing the unscheduled jobs accordingto the shortest processing time (SPT) rule according toLemma 3 Consequently the first lower bound is
LB1=
119896
sum
119894=1
119862[119894](119878) +
119899minus119896
sum
119895=1
119862(119895) (7)
where 119862(119895)
= 119862[119896](119878) + sum
119895
119894=1119901(119896+119894)
(1 + sum119896
119897=1119901[119897]
+
sum119894minus1
119897=1119901(119899minus119896+119897minus1)
)119886 On the other hand this lower bound may
not be tight if the release time is long To overcome thissituation a second lower bound is established by takingaccount of the release time The completion time for the(119896 + 1)th job is
119862[119896+1]
(119878) = max 119862[119896]
(119878) 119903[119896+1]
+ 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
ge 119903[119896+1]
(119878) + 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
(8)
Similarly the completion time for the (119896 + 119895)th job is
119862[119896+119895]
(119878) ge 119903[119896+119895]
(119878) + 119901[119896+119895]
(1 +
119896
sum
119897=1
119901[119897]
+
119895minus1
sum
119897=1
119901[119896+119897]
)
119886
1 le 119895 le 119899 minus 119896
(9)
Note that 119862[119896+119895]
(119878) is greater than or equal to 1199031015840
(119896+119895) where
1199031015840
(119896+1)le 1199031015840
(119896+2)le sdot sdot sdot le 119903
1015840
(119899)denote the release times of
the unscheduled jobs arranged in a nondecreasing orderThesecond term on the right hand side of (9) is minimized bythe SPT rule since (1 +sum
119896
119897=1119901[119897]+sum119895minus1
119897=1119901[119896+119897]
)119886 is a decreasing
function of sum119895minus1119897=1
119901[119896+119897]
It follows that we have the followingsecond lower bound
LB2=
119896
sum
119894=1
119862[119894]
+
119899minus119896
sum
119895=1
119862(119895) (10)
where 119862(119895)
= 1199031015840
(119896+119895)(119878) +119901
1015840
(119896+1)(1+sum
119896
119897=1119901[119897]+sum119895minus1
119897=11199011015840
(119899minus119896+119897minus1))119886
Note that 1199011015840(119896+119895)
and 1199031015840
(119896+119895)do not necessarily come from the
same job In order tomake the lower bound tighter we choosethe maximum value from (7) and (10) as the lower bounds of119875119878 That is
LB = max LB1 LB2 (11)
43TheProcedure of Genetic Algorithms Agenetic algorithm(GA) is an optimization method that mimics natural pro-cesses GAwas invented by Holland [40] and themost widelyused to solve numerical optimization problems in a widevariety of application fields including biology economicsengineering business agriculture telecommunications andmanufacturing For example in Goldberg [41] authors usingGA in engineering design problems is reviewed in Gen andCheng [42] Soolaki et al [43] use a GA to solve an airlineboarding problem with linear programming models [44 45]and use genetic algorithms to optimize the parameters for thegiven test collections GAs start evolving by generating aninitial population of chromosomes Then a fitness functionis used to compute the relative fitness of each chromosomeof the population The selection crossover and mutationoperators are used in succession to create a new population
Mathematical Problems in Engineering 5
of chromosomes for the next generation This approach hasgained increasing popularity in solving many combinatorialoptimization problems in a wide variety of different disci-plines
431 Initial Settings In a GA every problem is presented bya code and each code is seen as a geneThe existing genes canbe combined and seen as a chromosome each of which is oneof the feasible solutions to a problem However traditionalrepresentation of GA does not work for scheduling problems(Etiler et al [46]) In dealing with this condition this studyadopts the same method that a structure can describe thejobs as a sequence in the problem To specify our approachseveral initial sequences are adopted In GA
1 jobs are placed
according to the shortest processing times (SPT) first ruleIn GA
2 jobs are arranged in earliest ready times (ERT) first
rule In GA3 jobs are arranged in a nondecreasing order on
the sum of job processing times and ready times Note thatbefore performing GA NEH algorithm (Nawaz et al [47]) isutilized to improve the quality of the solutions obtained fromthe previous rules to reduce many idle periods The processof GA
1 GA2 and GA
3are different initial sequences and use
the same selection crossover mutation operators populationsize and generations to obtain near-optimal solution Inaddition the fourth genetic algorithm denoted as GA
4 is
the best one among GA1 GA2 and GA
3 that is GA
4=
minGA1GA2GA3
In order to avoid rapidly observing a local optimum in asmall population or consume more waiting time in a largeone this study set a suitable population size as 60 (119873 =
60) in a preliminary trial It is also an important work toevaluate the fitness of selected chromosomes that each ofthe chromosomes is included or excluded from a feasiblesolution The main goal of this study is to minimize thetotal completion time Assume that 119878
119894(119905) is the 119894th string
in the 119894th generation and the total completion time of 119878119894(119905)
is sum119899
119895=1119862119895(119878119894(119905)) Then the fitness function of 119878
119894(119905) can be
represented as 119891(119878119894(119905)) Following are the calculations of the
strings in fitness function
119891 (119878119894(119905)) = max
1le119897le119873
119899
sum
119895=1
119862119895(119878119897(119905))
minus
119899
sum
119895=1
119862119895(119878119894(119905))
(12)
Moreover it is also crucial work to ensure that the probabilityof selection for a sequence with lower value of the objectivefunction is higher Thus the probability 119875(119878
119894(119905)) can be
written as follows
119875 (119878119894(119905)) =
119891 (119878119894(119905))
sum119899
119894=1119891 (119878119894(119905))
(13)
432 Operators There are a few operators that are used inthis study Following are the descriptions of those operatorscrossover mutation and selection
(a) CrossoverThis is an operator that exchanges some of thegenes of the selected parents with the main concept being
005 010 015 020 025 030000
001
001
002
002
003
Mea
n er
ror (
)
00040 00192 00108 00103 0020600171Pm
Figure 1 The performance of the genetic algorithms for various 119875119898
at (119899 119886 120579119873 119892) = (20 minus005 05 60 100119899)
that the descendant can inherit the advantages of its parentsThis study applied the linear order crossover operator (LOX)proposed by Falkenauer and Bouffouix [48] and is one of thebetter performers among the others (Etiler et al [46]) Theprobability of crossover is set to 1
(b) Mutation The main object of mutation is to achieve foran overall optimal solution and to avoid a locally optimalone In this study the mutation rates (119875
119898) are set at 010
based on our preliminary experiment as shown in Figure 1For (119899 119886 120579119873 119892) = (20 minus005 05 60 100119899) 100 sets of datawere randomly generated to evaluate the performance of theproposed algorithms with varying values of 119875
119898 The results
showed that the proposed algorithmshad the leastmean errorpercentage at 119875
119898= 010
(c) Selection This is a process that determines the proba-bility of each chromosome and is used to decide the betterchromosomes with the better fitness value The evolutionimplemented in our algorithm is based on the elitist list Wecopy the best offspring and use them to generate some ofthe next generation The rest of the offspring are generatedfrom the parent chromosomes by the roulette wheel selectionmethod which can maintain the variety of genes
433 Stopping Criteria In the preliminary experimentthe proposed GAs are terminated after 100 lowast 119899 genera-tions as shown in Figures 2 and 3 For (119899 119886 120579119873 119875
119898) =
(20 minus005 05 60 010) the above 100 sets of randomly gen-erated data were used to evaluate the performance of theproposed algorithms with varying values of 119892 The resultsshowed that the least mean error percentage of the proposedalgorithms would stabilize with reasonable CPU time rangeafter 119892 = 100119899
5 Computational Experiment
A computational experiment was conducted to evaluatethe efficiency of the branch-and-bound algorithm and
6 Mathematical Problems in Engineering
00738 00510 00315 00258 00040
000
001
002
003
004
005
006
007
008
Mea
n er
ror (
)
40n 60n 80n 100n20n
g
Figure 2 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
01666 03260 04981 06800 08382
000
010
020
030
040
050
060
070
080
090
CPU
tim
es (s
)
g
40n 60n 80n 100n20n
Figure 3 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
the accuracies of the genetic algorithmsThe algorithms werecoded in Fortran and run on Compaq Visual Fortran version66 on an Intel(R)Core(TM)2QuadCPU266GHzwith 4GBRAM on Windows Vista The experimental design followedReeves [49] design The job processing times were generatedfrom a uniform distribution over the integers between 1 and20 in every case while the release times were generated froma uniform distribution over the integers on (0 20119899120579) where 119899is the number of jobs Five different sets of problem instanceswere generated by giving 120579 the values 1119899 025 05 075and 1
For the branch-and-bound algorithm the average and themaximum numbers of nodes as well as the average and themaximum execution times (in seconds) were recorded Forthe three genetic algorithms the mean and the maximum
0
200
400
600
800
1000
1200
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005
minus010
minus015
minus020
120579
1n
Figure 4Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 12
error percentages were recorded where the error percentagewas calculated as
(GA119894minus TClowast)
TClowastlowast 100 (14)
where GA119894(119894 = 1 2 3 4) is the total completion time
obtained from the genetic algorithmand TClowast is the totalcompletion time of the optimal schedule The computationaltimes of the heuristic algorithmswere not recorded since theywere finished within a second
In the computational experiment four different numbersof jobs (119899 = 12 16 20 and 24) four different values of learn-ing effect (119886 = minus005 minus010 minus015 andminus020) and five differ-ent values of generation parameter of release times (120579 = 1119899025 05 075 and 1) were tested in the branch-and-boundalgorithmAs a consequence 80 experimental situationswereexamined A set of 20 instances were randomly generatedfor each situation and a total of 1600 problems were testedThe algorithms were set to skip to the next set of data if thenumber of nodes exceeded 108 The results are presented inTable 1 and Figures 4 5 6 and 7 Figures 4ndash7 showed theaverage number of nodes for various 120579 and 119886 at job size 1216 20 and 24 respectively The average number of nodesdecreased as the value of 120579 increased when 119899 was greaterthan 16 This was the direct result of the efficiency of LB
1
and LB2 As 120579 increased the frequency of applications of LB
2
would increase Consequently it would yield longer releasetimes in those cases and the properties were more powerfulMoreover LB
1is more efficient than LB
2 Table 1 and Figures
4ndash7 also showed whether in job size the algorithms had theleast mean number of nodes at 120579 = 1119899 It was due to thefact that with 120579 = 1119899 the release time was relatively shortand the completion time would readily exceed the release
Mathematical Problems in Engineering 7
0
10000
20000
30000
40000
50000
60000
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005minus010
minus015
minus020
120579
1n
Figure 5The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 16
time In those cases Property 6 was applied more frequentlyconversely the completion time would not easily exceed therelease time when the values of 120579 increased Moreover thenumber of nodes increased exponentially as the number ofjobs increased which was typical of an NP-hard problem Asillustrated in Table 1 when 119899 = 24 there were five cases inwhich the branch-and-bound algorithm could solve all theproblems optimally larger than 10
8 nodes The branch-and-bound algorithm had the worst performance when (119899 119886 120579) =
(24 minus005 025)with 87times107 nodes and 5234 secondsWith
120579 fixed at 1119899 the decrease of the completion time would berelatively small at the beginning when the learning effectwassmall (eg 119886 = minus005) In other words the completion timewould easily exceed the release time which would expeditethe timing of invoking Property 6 and consequently theaverage number of nodes would be smaller With 120579 = 025 as119899 increased the corresponding least average number of nodeswould occur at greater values of learning effect
The performance of the proposed GA algorithms out ofthe 80 evaluations and a total of 1600 problems was testedThe number of times that each of the objective functions ofthe GA
1 GA2 and GA
3algorithms had the smallest mean
error percentage was 45 41 and 49 respectively In additionin Table 1 and Figures 8 9 and 10 their performanceswere not affected with the learning rate the generationparameter 120579 of release times or the number of jobs Noneof the three genetic algorithms had absolutely dominantperformance in terms of mean error percentage Howeverthe combined algorithm GA
4strikingly outperformed each
of the three algorithms in terms of the maximummean error
0
500000
1000000
1500000
2000000
2500000
3000000
3500000
4000000
4500000
5000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 6Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 20
0
5000000
10000000
15000000
20000000
25000000
30000000
35000000
40000000
45000000
50000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 7The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 24
8 Mathematical Problems in Engineering
Table1Th
eperform
ance
oftheb
ranch-
and-bo
undandgenetic
algorithm
s
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
12
minus005
111989942
385
00031
00156
2000858
17151
000
00000
00004
0108013
000
00000
00025
549
1781
00187
00624
2001055
21091
00142
02839
000
00000
00000
00000
00050
473
1189
00140
00312
20000
00000
00000
00000
00000
00000
00000
00000
00075
414
1013
00125
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0569
2427
00164
00624
20000
00000
00000
00000
00000
00000
00000
00000
00
minus010
111989962
488
00023
00156
2001646
20568
01617
20568
000
00000
00000
00000
00025
1047
4318
00312
01092
20000
4400878
000
4400878
00283
05665
000
00000
00050
480
1321
00140
004
6820
000
00000
00006
8413
676
00019
00373
000
00000
00075
786
5291
00203
01404
20000
00000
00000
00000
00000
00000
00000
00000
0010
0396
1382
00125
004
6820
000
00000
00000
00000
00000
00000
00000
00000
00
minus015
111989921
7100016
00156
2000206
04130
00000
00000
00360
04130
00000
00000
025
817
4363
00226
00936
20000
00000
00000
00000
0001228
24567
000
00000
00050
493
1969
00148
00624
20000
00000
00000
00000
00000
00000
00000
00000
00075
286
803
000
9400312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0417
1259
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
111989947
336
00023
00156
2001295
25896
01295
25896
000
03000
69000
00000
00025
1091
6689
00289
01560
2000106
02119
000
00000
00000
00000
00000
00000
00050
337
948
00109
00312
2000314
06284
000
00000
00000
00000
00000
00000
00075
452
1121
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0319
1387
00086
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
16
minus005
111989974
365
00078
00312
2002066
34518
02202
27743
01114
12163
006
0812
163
025
39325
105847
20108
58188
2004522
53071
01479
17548
00621
04629
00309
03126
050
18779
107488
08697
44616
2000052
01035
00935
16983
00849
16983
000
00000
00075
22967
290574
09259
102025
20000
00000
00000
00000
00000
00000
00000
00000
0010
05144
34260
02535
15288
20000
4100811
00115
02306
000
00000
00000
00000
00
minus010
111989991
323
00086
00312
2001813
34958
02368
27957
01906
19030
000
00000
00025
53237
291654
25826
127297
2003151
25393
02824
21606
01918
21606
01271
21606
050
15022
193968
06778
81277
2000616
10305
00154
03088
00154
03088
00000
00000
075
15144
197196
06474
80029
20000
00000
00000
00000
00000
0600120
000
00000
0010
03403
9875
01794
04992
20000
00000
00000
00000
00000
00000
00000
00000
00
minus015
1119899118
1368
00101
01092
2002462
45183
00431
06619
00103
02068
000
00000
00025
52185
4846
6222994
190165
2000762
08319
03212
37507
00969
10191
00418
08319
050
17808
101503
08221
46020
2000740
07668
00356
06773
00574
06773
00339
06773
075
3763
16453
01997
08112
20000
00000
00000
00000
00000
6701343
000
00000
0010
03882
14411
01911
07020
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899200
828
00172
00624
2001832
26781
03397
1766
003644
22420
000
00000
00025
25881
149925
11224
53664
2000140
02772
00299
04614
000
9801922
000
0200034
050
11829
48284
05819
23556
20000
00000
0000072
01431
000
00000
00000
00000
00075
2687
12382
01427
05928
20000
00000
00000
00000
00000
00000
00000
00000
0010
02047
5258
01076
02340
20000
00000
00000
00000
00000
00000
00000
00000
00
Mathematical Problems in Engineering 9
Table1Con
tinued
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
20
minus005
111989982
278
00156
00624
2016
268
74028
03538
69562
00113
02265
000
00000
00025
1319111
4407360
1059941
3224697
2007803
29510
05928
32815
08700
42205
02480
15709
050
963361
9983475
618895
5687017
2000377
040
1601608
1746
402224
15708
00033
00510
075
323776
2518001
229485
1831606
2000022
004
46000
00000
0000027
004
46000
00000
0010
041188
138955
33642
1110
7220
00017
00338
00017
00338
000
00000
00000
00000
00
minus010
11198991345
14123
01716
17317
2006707
34958
04690
28787
04353
34958
00000
00000
025
2134707
16978059
1543318
12044214
2003845
23175
01848
06798
03536
21979
004
6002909
050
202434
1320363
156266
1024927
2000000
00000
000
4500786
00000
00000
00000
00000
075
67733
450297
55645
3452
2920
000
00000
00000
00000
00000
00000
00000
00000
0010
030928
238389
26192
185483
2000023
004
62000
00000
0000023
004
62000
00000
00
minus015
1119899293
2224
004
2902964
2001670
13088
01845
29345
01835
29345
000
00000
00025
1539138
7544
291
1039052
4970
186
20046
8529738
01876
10582
02690
19134
00544
09353
050
235214
2737579
16906
61920840
20000
4600923
00032
006
4100073
01284
000
00000
00075
56901
281822
45716
225576
20000
00000
00000
00000
00000
00000
00000
00000
0010
017992
51529
15163
36973
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899698
4825
00991
06240
2002436
15781
04239
404
2602224
15781
01927
15781
025
434800
664
668017
2750735
40004590
2001596
14118
00549
02968
02765
21869
00203
02249
050
107201
675811
80878
476895
20000
00000
00000
00000
0000058
01155
000
00000
00075
31529
163854
25990
110918
20000
00000
00000
00000
00000
00000
00000
00000
0010
015118
63845
13198
49453
20000
00000
0000016
00324
00016
00318
000
00000
00
24
minus005
1119899478
2082
01248
04836
2004707
39003
02017
20990
01368
18118
00959
18118
025
43205517
875044
4452338760
1060
43535
1304991
21418
06032
19032
046
0720265
02283
13231
050
11780220
66191783
13054869
65948027
1900758
10456
00313
02363
01334
10456
00161
02363
075
1216867
4261994
1562529
4708125
2000056
01126
00198
03359
00174
01871
000
00000
0010
0888475
5439067
1144057
69244
1720
000
00000
00000
00000
00000
00000
00000
00000
00
minus010
11198991074
12050
02496
26364
2007378
42545
03868
24023
02577
18340
00541
06057
025
24953166
5300
6630
28547095
6344
0156
1104324
22382
05843
29043
06883
25478
03452
22382
050
3116322
31612798
3661781
38159717
2000137
0118
400087
00883
00075
01028
000
00000
00075
513685
2435858
672926
2964331
20000
00000
0000015
00306
00112
01371
000
00000
0010
0270431
1462391
35260
91960620
20000
00000
0000052
01037
00039
00771
000
00000
00
minus015
1119899674
2613
01693
06084
2007280
78202
03115
27017
02512
34255
00284
05683
025
23417410
61951876
253876
736371360
415
09250
33429
02948
12902
02232
17855
00499
04354
050
4118014
36622393
4749373
39276216
2000114
0119
300053
01058
000
00000
00000
00000
00075
661765
5958864
874869
7920327
2000073
01458
00075
01509
000
4300867
000
00000
0010
0837932
11995859
968626
13244172
2000000
00000
00000
00000
00000
00000
00000
00000
minus020
11198993126
15554
07082
31980
2001942
32396
04290
33146
03196
32396
01669
32396
025
9333827
72575949
9993
219
80682031
1500589
02822
006
4607914
00844
07331
00204
02057
050
384242
2271989
508384
2808486
2000038
00760
00130
02596
00130
02596
00000
00000
075
570054
4053320
7017
325311990
20000
00000
00000
00000
00000
00000
00000
00000
0010
0233727
1202819
2915
7413260
0620
000
00000
00000
00000
00000
00000
00000
00000
00NoteldquoO
Frdquodeno
testhe
numbero
finstances
in20
setsof
datathatcanbe
solved
inlessthan
108no
desb
yusingtheb
ranch-
and-bo
undmetho
d
10 Mathematical Problems in Engineering
000
005
010
015
020
025
030
GA1 GA2 GA3 GA4GA
n16
n20n12
n24
Mea
n er
ror (
)
Figure 8 The performance of the genetic algorithms for various 119899
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
Mea
n er
ror (
)
minus005
minus010 minus020
minus015
Figure 9 The performance of the genetic algorithms for various 119886
percentage among all the 80 scenarios with varying learningrates numbers of jobs and values of 120579 The maximummean error percentages of GA
1 GA2 GA3 and GA
4were
16268 06032 08700 and 03452 respectively Themaximum mean error percentage of GA
1was more than
four times that of GA4 The combined algorithm GA
4also
clearly outperformed each of the three algorithms in terms
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
02505
0751
1n
Mea
n er
ror (
)
Figure 10The performance of the genetic algorithms for various 120579
of the maximum error percentage The worst cases of thecorresponding algorithms were 78202 69562 42205and 32396 respectively The worst case error of GA
1was
more than twice that of GA4Thus we would recommend the
combined algorithm GA4
6 Conclusions
In this paper we address a single-machine total completiontime problem with the sum of processing times basedlearning effect and release timesThe objective is to minimizethe total completion time The problem without learningconsideration is NP-hard one Thus we develop a branch-and-bound algorithm incorporatingwith several dominancesand two lower bounds to derive optimal solution and geneticalgorithms that was proposed to obtain near-optimal solu-tions respectively
The branch-and-bound algorithm performs well to solvethe instances of less than or equal to 24 jobs in a reasonableamount of time In the different varying parameter theproposed genetic algorithms are quite accurate for small sizeproblems and the mean error percentage of the proposedgenetic algorithms are less than 035 Another interestingtopic for future study is to consider and study the problem inthe multimachine environments or multicriteria cases
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 3
job preemption are not allowed Each job 119895 has a normalprocessing time 119901
119895and a release time 119903
119895 The general job
learning model is 119901119895119903
= 119901119895(1 + sum
119903minus1
119897=1119901[119897])119886 where 119901
119895119903is
the actual processing time of job 119895 scheduled in the 119903thposition and 119886 le 0 is a learning ratio The objective of thisproblem is to find an optimal schedule 119878
lowast that minimizesthe total completion time that is sum119899
119894=1119862119894(119878lowast) le sum
119899
119894=1119862119894(119878)
for any schedule 119878 Using the standard three-field notation ofGraham et al [38] our scheduling problem can be denotedas 1119901
119895119903= 119901119895(1 + sum
119903minus1
119897=1119901[119897])119886sum119862
4 The Branch-and-Bound andGenetic Algorithms
In this paper we will apply the branch-and-bound andgenetic algorithms search for the optimal solution and obtainnear-optimal solution respectively First in order to facilitatethe searching process and improve the branching procedurewe develop some adjacent pairwise interchange propertiesand two lower bounds to use in branch-and-bound algo-rithmThen the procedure of the genetic algorithms is givenat last
41 Dominance Properties Before presenting the adjacentpairwise interchange properties we provide two lemmaswhich will be used in the proofs of the properties in thesequel
Lemma 1 Let119891(119909) = 1+ 119886119909(1 minus 119909)119886minus1
minus (1 + 119909)119886 then119891(119909) ge
0 for 119886 le 0 and 119909 gt 0
Lemma 2 Let 119892(120582) = (120582 minus 1) + (1 + 120582119909)119886minus 120582(1 + 119909)
119886 then119892(120582) ge 0 for 120582 ge 1 119886 le 0 and 119909 gt 0
To fathom the searching tree we develop some dom-inance properties based on a pairwise interchange of twoadjacent jobs 119869
119894and 119869
119895 Let 119878 = (120587 119869
119894 119869119895 1205871015840) and 119878
1015840=
(120587 119869119895 119869119894 1205871015840) be two sequences in which 120587 and 120587
1015840 denotepartial sequences To show that 119878 dominates 1198781015840 it suffices toshow that 119862
119894(119878) + 119862
119895(119878) le 119862
119895(1198781015840) + 119862119894(1198781015840) and 119862
119895(119878) lt 119862
119894(1198781015840)
In addition let 119905 be the completion time of the last job insubsequence 120587 with (119903 minus 1) jobs
Property 1 If 119901119894lt 119901119895and max119903
119894 119903119895 le 119905 then 119878 dominates
1198781015840
Proof Since max119903119894 119903119895 le 119905 we have
119862119894(119878) = 119905 + 119901
119894(1 +
119903minus1
sum
119897=1
119901[119897])
119886
119862119895(119878) = 119905 + 119901
119894(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119895(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119894)
119886
119862119895(1198781015840) = 119905 + 119901
119895(1 +
119903minus1
sum
119897=1
119901[119897])
119886
119862119894(1198781015840) = 119905 + 119901
119895(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119894(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119895)
119886
(1)
After taking the difference of (1) we have
119862119894(1198781015840) minus 119862119895(119878) = (119901
119895minus 119901119894)(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119894(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119895)
119886
minus 119901119895(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119894)
119886
(2)
On substituting120582 = 119901119895119901119894 119906 = (1+sum
119903minus1
119897=1119901[119897]) and119909 = (119901
119894(1+
sum119899
119897=1119901119897)) into (2) and simplifying it we obtain
119862119894(1198781015840) minus 119862119895(119878) = 119901
119894119906119886[(120582 minus 1) + (1 + 120582119909)
119886minus 120582(1 + 119909)
119886]
(3)
By Lemma 2 with 120582 gt 1 119886 le 0 and 119909 gt 0 we have 119862119894(1198781015840) minus
119862119895(119878) gt 0Moreover after taking the difference of total completion
times (TC) between sequences 119878 and 1198781015840 we have
TC (1198781015840) minus TC (119878)
= [119862119895(1198781015840) + 119862119894(1198781015840)] minus [119862
119894(119878) + 119862
119895(119878)]
= 2 (119901119895minus 119901119894)(1 +
119903minus1
sum
119897=1
119901[119897])
119886
+ 119901119894(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119895)
119886
minus 119901119895(1 +
119903minus1
sum
119897=1
119901[119897]
+ 119901119894)
119886
(4)
By (3) it can be easily shown that (4) is nonnegative for 119901119894lt
119901119895 Therefore 119878 dominates 1198781015840
The proofs of Properties 2 to 5 are omitted since they aresimilar to that of Property 1
Property 2 If 119903119894le 119905 le 119903
119895le 119905 + 119901
119894(1 + sum
119903minus1
119897=1119901[119897])119886
and 119901119894lt 119901119895
then 119878 dominates 1198781015840
Property 3 If 119905 ge 119903119894and 119905 + 119901
119894(1 + sum
119903minus1
119897=1119901[119897])119886
lt 119903119895 then 119878
dominates 1198781015840
Property 4 If 119905 le 119903119894le 119903119895 119903119894+ 119901119894(1 + sum
119903minus1
119897=1119901[119897])119886
ge 119903119895 and
119901119894lt 119901119895 then 119878 dominates 1198781015840
Property 5 If 119905 le 119903119894and 119903119894+ 119901119894(1 + sum
119903minus1
119897=1119901[119897])119886
lt 119903119895 then 119878
dominates 1198781015840
4 Mathematical Problems in Engineering
In order to further determine the ordering of the remain-ing unscheduled jobs to further speed up the searchingprocess we provide the following property Assume that 119878 =
(120587 120587119888) is a sequence of jobs where 120587 is the scheduled part
containing 119896 jobs and 120587119888 is the unscheduled part Let 119878
1=
(120587 1205871015840) be the sequence in which the unscheduled jobs are
arranged in a nondecreasing order of job processing timesthat is 119901
(119896+1)le 119901(119896+2)
le sdot sdot sdot le 119901(119899)
Property 6 If 119862[119896](1198781) gt max
119895isin120587119888119903119895 then 119878
1= (120587 120587
1015840)
dominates sequences of the type (120587 120587119888) for any unscheduledsequence 120587119888
Proof Since 119862[119896](1198781) gt max
119895isin120587119888119903119895 it implies that all the
unscheduled jobs are ready to be processed on time 119862[119896](1198781)
To obtain the optimal subsequence let 1198781= (120587 120587
1015840) be the
sequence in which the unscheduled jobs are arranged innondecreasing order of jobs processing times
42 Lower Bounds In this subsection we develop two lowerbounds by using the following lemma from Hardy et al [39]
Lemma 3 Suppose that 119886119894and 119887
119894are two sequences of
numbers The sum sum119899
119894=1119886119894119887119894of products of the corresponding
elements is the least if the sequences are monotonic in theopposite sense
First let 119875119878 be a partial schedule in which the order ofthe first 119896 jobs has been determined and let 119878 be a completeschedule obtained from 119875119878 By definition the completiontime for the (119896 + 1)th job is
119862[119896+1]
(119878) = max 119862[119896]
(119878) 119903[119896+1]
+ 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
ge 119862[119896]
(119878) + 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
(5)
Similarly the completion time for the (119896 + 119895)th job is
119862[119896+119895]
(119878) ge 119862[119896]
(119878) +
119895
sum
119894=1
119901(119896+119894)
(1 +
119896
sum
119897=1
119901[119897]
+
119894minus1
sum
119897=1
119901(119896+119897)
)
119886
1 le 119895 le 119899 minus 119896
(6)
The first term on the right hand side of (6) is known anda lower bound of the total completion time for the partialsequence 119875119878 can be obtained by minimizing the secondterm Since the value of (1 + sum
119896
119897=1119901[119897]
+ sum119894minus1
119897=1119901[119896+119897]
)119886 is a
decreasing function of sum119894minus1119897=1
119901[119896+119897]
the total completion timeis minimized by sequencing the unscheduled jobs accordingto the shortest processing time (SPT) rule according toLemma 3 Consequently the first lower bound is
LB1=
119896
sum
119894=1
119862[119894](119878) +
119899minus119896
sum
119895=1
119862(119895) (7)
where 119862(119895)
= 119862[119896](119878) + sum
119895
119894=1119901(119896+119894)
(1 + sum119896
119897=1119901[119897]
+
sum119894minus1
119897=1119901(119899minus119896+119897minus1)
)119886 On the other hand this lower bound may
not be tight if the release time is long To overcome thissituation a second lower bound is established by takingaccount of the release time The completion time for the(119896 + 1)th job is
119862[119896+1]
(119878) = max 119862[119896]
(119878) 119903[119896+1]
+ 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
ge 119903[119896+1]
(119878) + 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
(8)
Similarly the completion time for the (119896 + 119895)th job is
119862[119896+119895]
(119878) ge 119903[119896+119895]
(119878) + 119901[119896+119895]
(1 +
119896
sum
119897=1
119901[119897]
+
119895minus1
sum
119897=1
119901[119896+119897]
)
119886
1 le 119895 le 119899 minus 119896
(9)
Note that 119862[119896+119895]
(119878) is greater than or equal to 1199031015840
(119896+119895) where
1199031015840
(119896+1)le 1199031015840
(119896+2)le sdot sdot sdot le 119903
1015840
(119899)denote the release times of
the unscheduled jobs arranged in a nondecreasing orderThesecond term on the right hand side of (9) is minimized bythe SPT rule since (1 +sum
119896
119897=1119901[119897]+sum119895minus1
119897=1119901[119896+119897]
)119886 is a decreasing
function of sum119895minus1119897=1
119901[119896+119897]
It follows that we have the followingsecond lower bound
LB2=
119896
sum
119894=1
119862[119894]
+
119899minus119896
sum
119895=1
119862(119895) (10)
where 119862(119895)
= 1199031015840
(119896+119895)(119878) +119901
1015840
(119896+1)(1+sum
119896
119897=1119901[119897]+sum119895minus1
119897=11199011015840
(119899minus119896+119897minus1))119886
Note that 1199011015840(119896+119895)
and 1199031015840
(119896+119895)do not necessarily come from the
same job In order tomake the lower bound tighter we choosethe maximum value from (7) and (10) as the lower bounds of119875119878 That is
LB = max LB1 LB2 (11)
43TheProcedure of Genetic Algorithms Agenetic algorithm(GA) is an optimization method that mimics natural pro-cesses GAwas invented by Holland [40] and themost widelyused to solve numerical optimization problems in a widevariety of application fields including biology economicsengineering business agriculture telecommunications andmanufacturing For example in Goldberg [41] authors usingGA in engineering design problems is reviewed in Gen andCheng [42] Soolaki et al [43] use a GA to solve an airlineboarding problem with linear programming models [44 45]and use genetic algorithms to optimize the parameters for thegiven test collections GAs start evolving by generating aninitial population of chromosomes Then a fitness functionis used to compute the relative fitness of each chromosomeof the population The selection crossover and mutationoperators are used in succession to create a new population
Mathematical Problems in Engineering 5
of chromosomes for the next generation This approach hasgained increasing popularity in solving many combinatorialoptimization problems in a wide variety of different disci-plines
431 Initial Settings In a GA every problem is presented bya code and each code is seen as a geneThe existing genes canbe combined and seen as a chromosome each of which is oneof the feasible solutions to a problem However traditionalrepresentation of GA does not work for scheduling problems(Etiler et al [46]) In dealing with this condition this studyadopts the same method that a structure can describe thejobs as a sequence in the problem To specify our approachseveral initial sequences are adopted In GA
1 jobs are placed
according to the shortest processing times (SPT) first ruleIn GA
2 jobs are arranged in earliest ready times (ERT) first
rule In GA3 jobs are arranged in a nondecreasing order on
the sum of job processing times and ready times Note thatbefore performing GA NEH algorithm (Nawaz et al [47]) isutilized to improve the quality of the solutions obtained fromthe previous rules to reduce many idle periods The processof GA
1 GA2 and GA
3are different initial sequences and use
the same selection crossover mutation operators populationsize and generations to obtain near-optimal solution Inaddition the fourth genetic algorithm denoted as GA
4 is
the best one among GA1 GA2 and GA
3 that is GA
4=
minGA1GA2GA3
In order to avoid rapidly observing a local optimum in asmall population or consume more waiting time in a largeone this study set a suitable population size as 60 (119873 =
60) in a preliminary trial It is also an important work toevaluate the fitness of selected chromosomes that each ofthe chromosomes is included or excluded from a feasiblesolution The main goal of this study is to minimize thetotal completion time Assume that 119878
119894(119905) is the 119894th string
in the 119894th generation and the total completion time of 119878119894(119905)
is sum119899
119895=1119862119895(119878119894(119905)) Then the fitness function of 119878
119894(119905) can be
represented as 119891(119878119894(119905)) Following are the calculations of the
strings in fitness function
119891 (119878119894(119905)) = max
1le119897le119873
119899
sum
119895=1
119862119895(119878119897(119905))
minus
119899
sum
119895=1
119862119895(119878119894(119905))
(12)
Moreover it is also crucial work to ensure that the probabilityof selection for a sequence with lower value of the objectivefunction is higher Thus the probability 119875(119878
119894(119905)) can be
written as follows
119875 (119878119894(119905)) =
119891 (119878119894(119905))
sum119899
119894=1119891 (119878119894(119905))
(13)
432 Operators There are a few operators that are used inthis study Following are the descriptions of those operatorscrossover mutation and selection
(a) CrossoverThis is an operator that exchanges some of thegenes of the selected parents with the main concept being
005 010 015 020 025 030000
001
001
002
002
003
Mea
n er
ror (
)
00040 00192 00108 00103 0020600171Pm
Figure 1 The performance of the genetic algorithms for various 119875119898
at (119899 119886 120579119873 119892) = (20 minus005 05 60 100119899)
that the descendant can inherit the advantages of its parentsThis study applied the linear order crossover operator (LOX)proposed by Falkenauer and Bouffouix [48] and is one of thebetter performers among the others (Etiler et al [46]) Theprobability of crossover is set to 1
(b) Mutation The main object of mutation is to achieve foran overall optimal solution and to avoid a locally optimalone In this study the mutation rates (119875
119898) are set at 010
based on our preliminary experiment as shown in Figure 1For (119899 119886 120579119873 119892) = (20 minus005 05 60 100119899) 100 sets of datawere randomly generated to evaluate the performance of theproposed algorithms with varying values of 119875
119898 The results
showed that the proposed algorithmshad the leastmean errorpercentage at 119875
119898= 010
(c) Selection This is a process that determines the proba-bility of each chromosome and is used to decide the betterchromosomes with the better fitness value The evolutionimplemented in our algorithm is based on the elitist list Wecopy the best offspring and use them to generate some ofthe next generation The rest of the offspring are generatedfrom the parent chromosomes by the roulette wheel selectionmethod which can maintain the variety of genes
433 Stopping Criteria In the preliminary experimentthe proposed GAs are terminated after 100 lowast 119899 genera-tions as shown in Figures 2 and 3 For (119899 119886 120579119873 119875
119898) =
(20 minus005 05 60 010) the above 100 sets of randomly gen-erated data were used to evaluate the performance of theproposed algorithms with varying values of 119892 The resultsshowed that the least mean error percentage of the proposedalgorithms would stabilize with reasonable CPU time rangeafter 119892 = 100119899
5 Computational Experiment
A computational experiment was conducted to evaluatethe efficiency of the branch-and-bound algorithm and
6 Mathematical Problems in Engineering
00738 00510 00315 00258 00040
000
001
002
003
004
005
006
007
008
Mea
n er
ror (
)
40n 60n 80n 100n20n
g
Figure 2 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
01666 03260 04981 06800 08382
000
010
020
030
040
050
060
070
080
090
CPU
tim
es (s
)
g
40n 60n 80n 100n20n
Figure 3 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
the accuracies of the genetic algorithmsThe algorithms werecoded in Fortran and run on Compaq Visual Fortran version66 on an Intel(R)Core(TM)2QuadCPU266GHzwith 4GBRAM on Windows Vista The experimental design followedReeves [49] design The job processing times were generatedfrom a uniform distribution over the integers between 1 and20 in every case while the release times were generated froma uniform distribution over the integers on (0 20119899120579) where 119899is the number of jobs Five different sets of problem instanceswere generated by giving 120579 the values 1119899 025 05 075and 1
For the branch-and-bound algorithm the average and themaximum numbers of nodes as well as the average and themaximum execution times (in seconds) were recorded Forthe three genetic algorithms the mean and the maximum
0
200
400
600
800
1000
1200
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005
minus010
minus015
minus020
120579
1n
Figure 4Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 12
error percentages were recorded where the error percentagewas calculated as
(GA119894minus TClowast)
TClowastlowast 100 (14)
where GA119894(119894 = 1 2 3 4) is the total completion time
obtained from the genetic algorithmand TClowast is the totalcompletion time of the optimal schedule The computationaltimes of the heuristic algorithmswere not recorded since theywere finished within a second
In the computational experiment four different numbersof jobs (119899 = 12 16 20 and 24) four different values of learn-ing effect (119886 = minus005 minus010 minus015 andminus020) and five differ-ent values of generation parameter of release times (120579 = 1119899025 05 075 and 1) were tested in the branch-and-boundalgorithmAs a consequence 80 experimental situationswereexamined A set of 20 instances were randomly generatedfor each situation and a total of 1600 problems were testedThe algorithms were set to skip to the next set of data if thenumber of nodes exceeded 108 The results are presented inTable 1 and Figures 4 5 6 and 7 Figures 4ndash7 showed theaverage number of nodes for various 120579 and 119886 at job size 1216 20 and 24 respectively The average number of nodesdecreased as the value of 120579 increased when 119899 was greaterthan 16 This was the direct result of the efficiency of LB
1
and LB2 As 120579 increased the frequency of applications of LB
2
would increase Consequently it would yield longer releasetimes in those cases and the properties were more powerfulMoreover LB
1is more efficient than LB
2 Table 1 and Figures
4ndash7 also showed whether in job size the algorithms had theleast mean number of nodes at 120579 = 1119899 It was due to thefact that with 120579 = 1119899 the release time was relatively shortand the completion time would readily exceed the release
Mathematical Problems in Engineering 7
0
10000
20000
30000
40000
50000
60000
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005minus010
minus015
minus020
120579
1n
Figure 5The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 16
time In those cases Property 6 was applied more frequentlyconversely the completion time would not easily exceed therelease time when the values of 120579 increased Moreover thenumber of nodes increased exponentially as the number ofjobs increased which was typical of an NP-hard problem Asillustrated in Table 1 when 119899 = 24 there were five cases inwhich the branch-and-bound algorithm could solve all theproblems optimally larger than 10
8 nodes The branch-and-bound algorithm had the worst performance when (119899 119886 120579) =
(24 minus005 025)with 87times107 nodes and 5234 secondsWith
120579 fixed at 1119899 the decrease of the completion time would berelatively small at the beginning when the learning effectwassmall (eg 119886 = minus005) In other words the completion timewould easily exceed the release time which would expeditethe timing of invoking Property 6 and consequently theaverage number of nodes would be smaller With 120579 = 025 as119899 increased the corresponding least average number of nodeswould occur at greater values of learning effect
The performance of the proposed GA algorithms out ofthe 80 evaluations and a total of 1600 problems was testedThe number of times that each of the objective functions ofthe GA
1 GA2 and GA
3algorithms had the smallest mean
error percentage was 45 41 and 49 respectively In additionin Table 1 and Figures 8 9 and 10 their performanceswere not affected with the learning rate the generationparameter 120579 of release times or the number of jobs Noneof the three genetic algorithms had absolutely dominantperformance in terms of mean error percentage Howeverthe combined algorithm GA
4strikingly outperformed each
of the three algorithms in terms of the maximummean error
0
500000
1000000
1500000
2000000
2500000
3000000
3500000
4000000
4500000
5000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 6Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 20
0
5000000
10000000
15000000
20000000
25000000
30000000
35000000
40000000
45000000
50000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 7The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 24
8 Mathematical Problems in Engineering
Table1Th
eperform
ance
oftheb
ranch-
and-bo
undandgenetic
algorithm
s
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
12
minus005
111989942
385
00031
00156
2000858
17151
000
00000
00004
0108013
000
00000
00025
549
1781
00187
00624
2001055
21091
00142
02839
000
00000
00000
00000
00050
473
1189
00140
00312
20000
00000
00000
00000
00000
00000
00000
00000
00075
414
1013
00125
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0569
2427
00164
00624
20000
00000
00000
00000
00000
00000
00000
00000
00
minus010
111989962
488
00023
00156
2001646
20568
01617
20568
000
00000
00000
00000
00025
1047
4318
00312
01092
20000
4400878
000
4400878
00283
05665
000
00000
00050
480
1321
00140
004
6820
000
00000
00006
8413
676
00019
00373
000
00000
00075
786
5291
00203
01404
20000
00000
00000
00000
00000
00000
00000
00000
0010
0396
1382
00125
004
6820
000
00000
00000
00000
00000
00000
00000
00000
00
minus015
111989921
7100016
00156
2000206
04130
00000
00000
00360
04130
00000
00000
025
817
4363
00226
00936
20000
00000
00000
00000
0001228
24567
000
00000
00050
493
1969
00148
00624
20000
00000
00000
00000
00000
00000
00000
00000
00075
286
803
000
9400312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0417
1259
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
111989947
336
00023
00156
2001295
25896
01295
25896
000
03000
69000
00000
00025
1091
6689
00289
01560
2000106
02119
000
00000
00000
00000
00000
00000
00050
337
948
00109
00312
2000314
06284
000
00000
00000
00000
00000
00000
00075
452
1121
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0319
1387
00086
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
16
minus005
111989974
365
00078
00312
2002066
34518
02202
27743
01114
12163
006
0812
163
025
39325
105847
20108
58188
2004522
53071
01479
17548
00621
04629
00309
03126
050
18779
107488
08697
44616
2000052
01035
00935
16983
00849
16983
000
00000
00075
22967
290574
09259
102025
20000
00000
00000
00000
00000
00000
00000
00000
0010
05144
34260
02535
15288
20000
4100811
00115
02306
000
00000
00000
00000
00
minus010
111989991
323
00086
00312
2001813
34958
02368
27957
01906
19030
000
00000
00025
53237
291654
25826
127297
2003151
25393
02824
21606
01918
21606
01271
21606
050
15022
193968
06778
81277
2000616
10305
00154
03088
00154
03088
00000
00000
075
15144
197196
06474
80029
20000
00000
00000
00000
00000
0600120
000
00000
0010
03403
9875
01794
04992
20000
00000
00000
00000
00000
00000
00000
00000
00
minus015
1119899118
1368
00101
01092
2002462
45183
00431
06619
00103
02068
000
00000
00025
52185
4846
6222994
190165
2000762
08319
03212
37507
00969
10191
00418
08319
050
17808
101503
08221
46020
2000740
07668
00356
06773
00574
06773
00339
06773
075
3763
16453
01997
08112
20000
00000
00000
00000
00000
6701343
000
00000
0010
03882
14411
01911
07020
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899200
828
00172
00624
2001832
26781
03397
1766
003644
22420
000
00000
00025
25881
149925
11224
53664
2000140
02772
00299
04614
000
9801922
000
0200034
050
11829
48284
05819
23556
20000
00000
0000072
01431
000
00000
00000
00000
00075
2687
12382
01427
05928
20000
00000
00000
00000
00000
00000
00000
00000
0010
02047
5258
01076
02340
20000
00000
00000
00000
00000
00000
00000
00000
00
Mathematical Problems in Engineering 9
Table1Con
tinued
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
20
minus005
111989982
278
00156
00624
2016
268
74028
03538
69562
00113
02265
000
00000
00025
1319111
4407360
1059941
3224697
2007803
29510
05928
32815
08700
42205
02480
15709
050
963361
9983475
618895
5687017
2000377
040
1601608
1746
402224
15708
00033
00510
075
323776
2518001
229485
1831606
2000022
004
46000
00000
0000027
004
46000
00000
0010
041188
138955
33642
1110
7220
00017
00338
00017
00338
000
00000
00000
00000
00
minus010
11198991345
14123
01716
17317
2006707
34958
04690
28787
04353
34958
00000
00000
025
2134707
16978059
1543318
12044214
2003845
23175
01848
06798
03536
21979
004
6002909
050
202434
1320363
156266
1024927
2000000
00000
000
4500786
00000
00000
00000
00000
075
67733
450297
55645
3452
2920
000
00000
00000
00000
00000
00000
00000
00000
0010
030928
238389
26192
185483
2000023
004
62000
00000
0000023
004
62000
00000
00
minus015
1119899293
2224
004
2902964
2001670
13088
01845
29345
01835
29345
000
00000
00025
1539138
7544
291
1039052
4970
186
20046
8529738
01876
10582
02690
19134
00544
09353
050
235214
2737579
16906
61920840
20000
4600923
00032
006
4100073
01284
000
00000
00075
56901
281822
45716
225576
20000
00000
00000
00000
00000
00000
00000
00000
0010
017992
51529
15163
36973
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899698
4825
00991
06240
2002436
15781
04239
404
2602224
15781
01927
15781
025
434800
664
668017
2750735
40004590
2001596
14118
00549
02968
02765
21869
00203
02249
050
107201
675811
80878
476895
20000
00000
00000
00000
0000058
01155
000
00000
00075
31529
163854
25990
110918
20000
00000
00000
00000
00000
00000
00000
00000
0010
015118
63845
13198
49453
20000
00000
0000016
00324
00016
00318
000
00000
00
24
minus005
1119899478
2082
01248
04836
2004707
39003
02017
20990
01368
18118
00959
18118
025
43205517
875044
4452338760
1060
43535
1304991
21418
06032
19032
046
0720265
02283
13231
050
11780220
66191783
13054869
65948027
1900758
10456
00313
02363
01334
10456
00161
02363
075
1216867
4261994
1562529
4708125
2000056
01126
00198
03359
00174
01871
000
00000
0010
0888475
5439067
1144057
69244
1720
000
00000
00000
00000
00000
00000
00000
00000
00
minus010
11198991074
12050
02496
26364
2007378
42545
03868
24023
02577
18340
00541
06057
025
24953166
5300
6630
28547095
6344
0156
1104324
22382
05843
29043
06883
25478
03452
22382
050
3116322
31612798
3661781
38159717
2000137
0118
400087
00883
00075
01028
000
00000
00075
513685
2435858
672926
2964331
20000
00000
0000015
00306
00112
01371
000
00000
0010
0270431
1462391
35260
91960620
20000
00000
0000052
01037
00039
00771
000
00000
00
minus015
1119899674
2613
01693
06084
2007280
78202
03115
27017
02512
34255
00284
05683
025
23417410
61951876
253876
736371360
415
09250
33429
02948
12902
02232
17855
00499
04354
050
4118014
36622393
4749373
39276216
2000114
0119
300053
01058
000
00000
00000
00000
00075
661765
5958864
874869
7920327
2000073
01458
00075
01509
000
4300867
000
00000
0010
0837932
11995859
968626
13244172
2000000
00000
00000
00000
00000
00000
00000
00000
minus020
11198993126
15554
07082
31980
2001942
32396
04290
33146
03196
32396
01669
32396
025
9333827
72575949
9993
219
80682031
1500589
02822
006
4607914
00844
07331
00204
02057
050
384242
2271989
508384
2808486
2000038
00760
00130
02596
00130
02596
00000
00000
075
570054
4053320
7017
325311990
20000
00000
00000
00000
00000
00000
00000
00000
0010
0233727
1202819
2915
7413260
0620
000
00000
00000
00000
00000
00000
00000
00000
00NoteldquoO
Frdquodeno
testhe
numbero
finstances
in20
setsof
datathatcanbe
solved
inlessthan
108no
desb
yusingtheb
ranch-
and-bo
undmetho
d
10 Mathematical Problems in Engineering
000
005
010
015
020
025
030
GA1 GA2 GA3 GA4GA
n16
n20n12
n24
Mea
n er
ror (
)
Figure 8 The performance of the genetic algorithms for various 119899
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
Mea
n er
ror (
)
minus005
minus010 minus020
minus015
Figure 9 The performance of the genetic algorithms for various 119886
percentage among all the 80 scenarios with varying learningrates numbers of jobs and values of 120579 The maximummean error percentages of GA
1 GA2 GA3 and GA
4were
16268 06032 08700 and 03452 respectively Themaximum mean error percentage of GA
1was more than
four times that of GA4 The combined algorithm GA
4also
clearly outperformed each of the three algorithms in terms
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
02505
0751
1n
Mea
n er
ror (
)
Figure 10The performance of the genetic algorithms for various 120579
of the maximum error percentage The worst cases of thecorresponding algorithms were 78202 69562 42205and 32396 respectively The worst case error of GA
1was
more than twice that of GA4Thus we would recommend the
combined algorithm GA4
6 Conclusions
In this paper we address a single-machine total completiontime problem with the sum of processing times basedlearning effect and release timesThe objective is to minimizethe total completion time The problem without learningconsideration is NP-hard one Thus we develop a branch-and-bound algorithm incorporatingwith several dominancesand two lower bounds to derive optimal solution and geneticalgorithms that was proposed to obtain near-optimal solu-tions respectively
The branch-and-bound algorithm performs well to solvethe instances of less than or equal to 24 jobs in a reasonableamount of time In the different varying parameter theproposed genetic algorithms are quite accurate for small sizeproblems and the mean error percentage of the proposedgenetic algorithms are less than 035 Another interestingtopic for future study is to consider and study the problem inthe multimachine environments or multicriteria cases
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
4 Mathematical Problems in Engineering
In order to further determine the ordering of the remain-ing unscheduled jobs to further speed up the searchingprocess we provide the following property Assume that 119878 =
(120587 120587119888) is a sequence of jobs where 120587 is the scheduled part
containing 119896 jobs and 120587119888 is the unscheduled part Let 119878
1=
(120587 1205871015840) be the sequence in which the unscheduled jobs are
arranged in a nondecreasing order of job processing timesthat is 119901
(119896+1)le 119901(119896+2)
le sdot sdot sdot le 119901(119899)
Property 6 If 119862[119896](1198781) gt max
119895isin120587119888119903119895 then 119878
1= (120587 120587
1015840)
dominates sequences of the type (120587 120587119888) for any unscheduledsequence 120587119888
Proof Since 119862[119896](1198781) gt max
119895isin120587119888119903119895 it implies that all the
unscheduled jobs are ready to be processed on time 119862[119896](1198781)
To obtain the optimal subsequence let 1198781= (120587 120587
1015840) be the
sequence in which the unscheduled jobs are arranged innondecreasing order of jobs processing times
42 Lower Bounds In this subsection we develop two lowerbounds by using the following lemma from Hardy et al [39]
Lemma 3 Suppose that 119886119894and 119887
119894are two sequences of
numbers The sum sum119899
119894=1119886119894119887119894of products of the corresponding
elements is the least if the sequences are monotonic in theopposite sense
First let 119875119878 be a partial schedule in which the order ofthe first 119896 jobs has been determined and let 119878 be a completeschedule obtained from 119875119878 By definition the completiontime for the (119896 + 1)th job is
119862[119896+1]
(119878) = max 119862[119896]
(119878) 119903[119896+1]
+ 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
ge 119862[119896]
(119878) + 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
(5)
Similarly the completion time for the (119896 + 119895)th job is
119862[119896+119895]
(119878) ge 119862[119896]
(119878) +
119895
sum
119894=1
119901(119896+119894)
(1 +
119896
sum
119897=1
119901[119897]
+
119894minus1
sum
119897=1
119901(119896+119897)
)
119886
1 le 119895 le 119899 minus 119896
(6)
The first term on the right hand side of (6) is known anda lower bound of the total completion time for the partialsequence 119875119878 can be obtained by minimizing the secondterm Since the value of (1 + sum
119896
119897=1119901[119897]
+ sum119894minus1
119897=1119901[119896+119897]
)119886 is a
decreasing function of sum119894minus1119897=1
119901[119896+119897]
the total completion timeis minimized by sequencing the unscheduled jobs accordingto the shortest processing time (SPT) rule according toLemma 3 Consequently the first lower bound is
LB1=
119896
sum
119894=1
119862[119894](119878) +
119899minus119896
sum
119895=1
119862(119895) (7)
where 119862(119895)
= 119862[119896](119878) + sum
119895
119894=1119901(119896+119894)
(1 + sum119896
119897=1119901[119897]
+
sum119894minus1
119897=1119901(119899minus119896+119897minus1)
)119886 On the other hand this lower bound may
not be tight if the release time is long To overcome thissituation a second lower bound is established by takingaccount of the release time The completion time for the(119896 + 1)th job is
119862[119896+1]
(119878) = max 119862[119896]
(119878) 119903[119896+1]
+ 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
ge 119903[119896+1]
(119878) + 119901[119896+1]
(1 +
119896
sum
119897=1
119901[119897])
119886
(8)
Similarly the completion time for the (119896 + 119895)th job is
119862[119896+119895]
(119878) ge 119903[119896+119895]
(119878) + 119901[119896+119895]
(1 +
119896
sum
119897=1
119901[119897]
+
119895minus1
sum
119897=1
119901[119896+119897]
)
119886
1 le 119895 le 119899 minus 119896
(9)
Note that 119862[119896+119895]
(119878) is greater than or equal to 1199031015840
(119896+119895) where
1199031015840
(119896+1)le 1199031015840
(119896+2)le sdot sdot sdot le 119903
1015840
(119899)denote the release times of
the unscheduled jobs arranged in a nondecreasing orderThesecond term on the right hand side of (9) is minimized bythe SPT rule since (1 +sum
119896
119897=1119901[119897]+sum119895minus1
119897=1119901[119896+119897]
)119886 is a decreasing
function of sum119895minus1119897=1
119901[119896+119897]
It follows that we have the followingsecond lower bound
LB2=
119896
sum
119894=1
119862[119894]
+
119899minus119896
sum
119895=1
119862(119895) (10)
where 119862(119895)
= 1199031015840
(119896+119895)(119878) +119901
1015840
(119896+1)(1+sum
119896
119897=1119901[119897]+sum119895minus1
119897=11199011015840
(119899minus119896+119897minus1))119886
Note that 1199011015840(119896+119895)
and 1199031015840
(119896+119895)do not necessarily come from the
same job In order tomake the lower bound tighter we choosethe maximum value from (7) and (10) as the lower bounds of119875119878 That is
LB = max LB1 LB2 (11)
43TheProcedure of Genetic Algorithms Agenetic algorithm(GA) is an optimization method that mimics natural pro-cesses GAwas invented by Holland [40] and themost widelyused to solve numerical optimization problems in a widevariety of application fields including biology economicsengineering business agriculture telecommunications andmanufacturing For example in Goldberg [41] authors usingGA in engineering design problems is reviewed in Gen andCheng [42] Soolaki et al [43] use a GA to solve an airlineboarding problem with linear programming models [44 45]and use genetic algorithms to optimize the parameters for thegiven test collections GAs start evolving by generating aninitial population of chromosomes Then a fitness functionis used to compute the relative fitness of each chromosomeof the population The selection crossover and mutationoperators are used in succession to create a new population
Mathematical Problems in Engineering 5
of chromosomes for the next generation This approach hasgained increasing popularity in solving many combinatorialoptimization problems in a wide variety of different disci-plines
431 Initial Settings In a GA every problem is presented bya code and each code is seen as a geneThe existing genes canbe combined and seen as a chromosome each of which is oneof the feasible solutions to a problem However traditionalrepresentation of GA does not work for scheduling problems(Etiler et al [46]) In dealing with this condition this studyadopts the same method that a structure can describe thejobs as a sequence in the problem To specify our approachseveral initial sequences are adopted In GA
1 jobs are placed
according to the shortest processing times (SPT) first ruleIn GA
2 jobs are arranged in earliest ready times (ERT) first
rule In GA3 jobs are arranged in a nondecreasing order on
the sum of job processing times and ready times Note thatbefore performing GA NEH algorithm (Nawaz et al [47]) isutilized to improve the quality of the solutions obtained fromthe previous rules to reduce many idle periods The processof GA
1 GA2 and GA
3are different initial sequences and use
the same selection crossover mutation operators populationsize and generations to obtain near-optimal solution Inaddition the fourth genetic algorithm denoted as GA
4 is
the best one among GA1 GA2 and GA
3 that is GA
4=
minGA1GA2GA3
In order to avoid rapidly observing a local optimum in asmall population or consume more waiting time in a largeone this study set a suitable population size as 60 (119873 =
60) in a preliminary trial It is also an important work toevaluate the fitness of selected chromosomes that each ofthe chromosomes is included or excluded from a feasiblesolution The main goal of this study is to minimize thetotal completion time Assume that 119878
119894(119905) is the 119894th string
in the 119894th generation and the total completion time of 119878119894(119905)
is sum119899
119895=1119862119895(119878119894(119905)) Then the fitness function of 119878
119894(119905) can be
represented as 119891(119878119894(119905)) Following are the calculations of the
strings in fitness function
119891 (119878119894(119905)) = max
1le119897le119873
119899
sum
119895=1
119862119895(119878119897(119905))
minus
119899
sum
119895=1
119862119895(119878119894(119905))
(12)
Moreover it is also crucial work to ensure that the probabilityof selection for a sequence with lower value of the objectivefunction is higher Thus the probability 119875(119878
119894(119905)) can be
written as follows
119875 (119878119894(119905)) =
119891 (119878119894(119905))
sum119899
119894=1119891 (119878119894(119905))
(13)
432 Operators There are a few operators that are used inthis study Following are the descriptions of those operatorscrossover mutation and selection
(a) CrossoverThis is an operator that exchanges some of thegenes of the selected parents with the main concept being
005 010 015 020 025 030000
001
001
002
002
003
Mea
n er
ror (
)
00040 00192 00108 00103 0020600171Pm
Figure 1 The performance of the genetic algorithms for various 119875119898
at (119899 119886 120579119873 119892) = (20 minus005 05 60 100119899)
that the descendant can inherit the advantages of its parentsThis study applied the linear order crossover operator (LOX)proposed by Falkenauer and Bouffouix [48] and is one of thebetter performers among the others (Etiler et al [46]) Theprobability of crossover is set to 1
(b) Mutation The main object of mutation is to achieve foran overall optimal solution and to avoid a locally optimalone In this study the mutation rates (119875
119898) are set at 010
based on our preliminary experiment as shown in Figure 1For (119899 119886 120579119873 119892) = (20 minus005 05 60 100119899) 100 sets of datawere randomly generated to evaluate the performance of theproposed algorithms with varying values of 119875
119898 The results
showed that the proposed algorithmshad the leastmean errorpercentage at 119875
119898= 010
(c) Selection This is a process that determines the proba-bility of each chromosome and is used to decide the betterchromosomes with the better fitness value The evolutionimplemented in our algorithm is based on the elitist list Wecopy the best offspring and use them to generate some ofthe next generation The rest of the offspring are generatedfrom the parent chromosomes by the roulette wheel selectionmethod which can maintain the variety of genes
433 Stopping Criteria In the preliminary experimentthe proposed GAs are terminated after 100 lowast 119899 genera-tions as shown in Figures 2 and 3 For (119899 119886 120579119873 119875
119898) =
(20 minus005 05 60 010) the above 100 sets of randomly gen-erated data were used to evaluate the performance of theproposed algorithms with varying values of 119892 The resultsshowed that the least mean error percentage of the proposedalgorithms would stabilize with reasonable CPU time rangeafter 119892 = 100119899
5 Computational Experiment
A computational experiment was conducted to evaluatethe efficiency of the branch-and-bound algorithm and
6 Mathematical Problems in Engineering
00738 00510 00315 00258 00040
000
001
002
003
004
005
006
007
008
Mea
n er
ror (
)
40n 60n 80n 100n20n
g
Figure 2 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
01666 03260 04981 06800 08382
000
010
020
030
040
050
060
070
080
090
CPU
tim
es (s
)
g
40n 60n 80n 100n20n
Figure 3 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
the accuracies of the genetic algorithmsThe algorithms werecoded in Fortran and run on Compaq Visual Fortran version66 on an Intel(R)Core(TM)2QuadCPU266GHzwith 4GBRAM on Windows Vista The experimental design followedReeves [49] design The job processing times were generatedfrom a uniform distribution over the integers between 1 and20 in every case while the release times were generated froma uniform distribution over the integers on (0 20119899120579) where 119899is the number of jobs Five different sets of problem instanceswere generated by giving 120579 the values 1119899 025 05 075and 1
For the branch-and-bound algorithm the average and themaximum numbers of nodes as well as the average and themaximum execution times (in seconds) were recorded Forthe three genetic algorithms the mean and the maximum
0
200
400
600
800
1000
1200
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005
minus010
minus015
minus020
120579
1n
Figure 4Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 12
error percentages were recorded where the error percentagewas calculated as
(GA119894minus TClowast)
TClowastlowast 100 (14)
where GA119894(119894 = 1 2 3 4) is the total completion time
obtained from the genetic algorithmand TClowast is the totalcompletion time of the optimal schedule The computationaltimes of the heuristic algorithmswere not recorded since theywere finished within a second
In the computational experiment four different numbersof jobs (119899 = 12 16 20 and 24) four different values of learn-ing effect (119886 = minus005 minus010 minus015 andminus020) and five differ-ent values of generation parameter of release times (120579 = 1119899025 05 075 and 1) were tested in the branch-and-boundalgorithmAs a consequence 80 experimental situationswereexamined A set of 20 instances were randomly generatedfor each situation and a total of 1600 problems were testedThe algorithms were set to skip to the next set of data if thenumber of nodes exceeded 108 The results are presented inTable 1 and Figures 4 5 6 and 7 Figures 4ndash7 showed theaverage number of nodes for various 120579 and 119886 at job size 1216 20 and 24 respectively The average number of nodesdecreased as the value of 120579 increased when 119899 was greaterthan 16 This was the direct result of the efficiency of LB
1
and LB2 As 120579 increased the frequency of applications of LB
2
would increase Consequently it would yield longer releasetimes in those cases and the properties were more powerfulMoreover LB
1is more efficient than LB
2 Table 1 and Figures
4ndash7 also showed whether in job size the algorithms had theleast mean number of nodes at 120579 = 1119899 It was due to thefact that with 120579 = 1119899 the release time was relatively shortand the completion time would readily exceed the release
Mathematical Problems in Engineering 7
0
10000
20000
30000
40000
50000
60000
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005minus010
minus015
minus020
120579
1n
Figure 5The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 16
time In those cases Property 6 was applied more frequentlyconversely the completion time would not easily exceed therelease time when the values of 120579 increased Moreover thenumber of nodes increased exponentially as the number ofjobs increased which was typical of an NP-hard problem Asillustrated in Table 1 when 119899 = 24 there were five cases inwhich the branch-and-bound algorithm could solve all theproblems optimally larger than 10
8 nodes The branch-and-bound algorithm had the worst performance when (119899 119886 120579) =
(24 minus005 025)with 87times107 nodes and 5234 secondsWith
120579 fixed at 1119899 the decrease of the completion time would berelatively small at the beginning when the learning effectwassmall (eg 119886 = minus005) In other words the completion timewould easily exceed the release time which would expeditethe timing of invoking Property 6 and consequently theaverage number of nodes would be smaller With 120579 = 025 as119899 increased the corresponding least average number of nodeswould occur at greater values of learning effect
The performance of the proposed GA algorithms out ofthe 80 evaluations and a total of 1600 problems was testedThe number of times that each of the objective functions ofthe GA
1 GA2 and GA
3algorithms had the smallest mean
error percentage was 45 41 and 49 respectively In additionin Table 1 and Figures 8 9 and 10 their performanceswere not affected with the learning rate the generationparameter 120579 of release times or the number of jobs Noneof the three genetic algorithms had absolutely dominantperformance in terms of mean error percentage Howeverthe combined algorithm GA
4strikingly outperformed each
of the three algorithms in terms of the maximummean error
0
500000
1000000
1500000
2000000
2500000
3000000
3500000
4000000
4500000
5000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 6Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 20
0
5000000
10000000
15000000
20000000
25000000
30000000
35000000
40000000
45000000
50000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 7The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 24
8 Mathematical Problems in Engineering
Table1Th
eperform
ance
oftheb
ranch-
and-bo
undandgenetic
algorithm
s
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
12
minus005
111989942
385
00031
00156
2000858
17151
000
00000
00004
0108013
000
00000
00025
549
1781
00187
00624
2001055
21091
00142
02839
000
00000
00000
00000
00050
473
1189
00140
00312
20000
00000
00000
00000
00000
00000
00000
00000
00075
414
1013
00125
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0569
2427
00164
00624
20000
00000
00000
00000
00000
00000
00000
00000
00
minus010
111989962
488
00023
00156
2001646
20568
01617
20568
000
00000
00000
00000
00025
1047
4318
00312
01092
20000
4400878
000
4400878
00283
05665
000
00000
00050
480
1321
00140
004
6820
000
00000
00006
8413
676
00019
00373
000
00000
00075
786
5291
00203
01404
20000
00000
00000
00000
00000
00000
00000
00000
0010
0396
1382
00125
004
6820
000
00000
00000
00000
00000
00000
00000
00000
00
minus015
111989921
7100016
00156
2000206
04130
00000
00000
00360
04130
00000
00000
025
817
4363
00226
00936
20000
00000
00000
00000
0001228
24567
000
00000
00050
493
1969
00148
00624
20000
00000
00000
00000
00000
00000
00000
00000
00075
286
803
000
9400312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0417
1259
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
111989947
336
00023
00156
2001295
25896
01295
25896
000
03000
69000
00000
00025
1091
6689
00289
01560
2000106
02119
000
00000
00000
00000
00000
00000
00050
337
948
00109
00312
2000314
06284
000
00000
00000
00000
00000
00000
00075
452
1121
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0319
1387
00086
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
16
minus005
111989974
365
00078
00312
2002066
34518
02202
27743
01114
12163
006
0812
163
025
39325
105847
20108
58188
2004522
53071
01479
17548
00621
04629
00309
03126
050
18779
107488
08697
44616
2000052
01035
00935
16983
00849
16983
000
00000
00075
22967
290574
09259
102025
20000
00000
00000
00000
00000
00000
00000
00000
0010
05144
34260
02535
15288
20000
4100811
00115
02306
000
00000
00000
00000
00
minus010
111989991
323
00086
00312
2001813
34958
02368
27957
01906
19030
000
00000
00025
53237
291654
25826
127297
2003151
25393
02824
21606
01918
21606
01271
21606
050
15022
193968
06778
81277
2000616
10305
00154
03088
00154
03088
00000
00000
075
15144
197196
06474
80029
20000
00000
00000
00000
00000
0600120
000
00000
0010
03403
9875
01794
04992
20000
00000
00000
00000
00000
00000
00000
00000
00
minus015
1119899118
1368
00101
01092
2002462
45183
00431
06619
00103
02068
000
00000
00025
52185
4846
6222994
190165
2000762
08319
03212
37507
00969
10191
00418
08319
050
17808
101503
08221
46020
2000740
07668
00356
06773
00574
06773
00339
06773
075
3763
16453
01997
08112
20000
00000
00000
00000
00000
6701343
000
00000
0010
03882
14411
01911
07020
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899200
828
00172
00624
2001832
26781
03397
1766
003644
22420
000
00000
00025
25881
149925
11224
53664
2000140
02772
00299
04614
000
9801922
000
0200034
050
11829
48284
05819
23556
20000
00000
0000072
01431
000
00000
00000
00000
00075
2687
12382
01427
05928
20000
00000
00000
00000
00000
00000
00000
00000
0010
02047
5258
01076
02340
20000
00000
00000
00000
00000
00000
00000
00000
00
Mathematical Problems in Engineering 9
Table1Con
tinued
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
20
minus005
111989982
278
00156
00624
2016
268
74028
03538
69562
00113
02265
000
00000
00025
1319111
4407360
1059941
3224697
2007803
29510
05928
32815
08700
42205
02480
15709
050
963361
9983475
618895
5687017
2000377
040
1601608
1746
402224
15708
00033
00510
075
323776
2518001
229485
1831606
2000022
004
46000
00000
0000027
004
46000
00000
0010
041188
138955
33642
1110
7220
00017
00338
00017
00338
000
00000
00000
00000
00
minus010
11198991345
14123
01716
17317
2006707
34958
04690
28787
04353
34958
00000
00000
025
2134707
16978059
1543318
12044214
2003845
23175
01848
06798
03536
21979
004
6002909
050
202434
1320363
156266
1024927
2000000
00000
000
4500786
00000
00000
00000
00000
075
67733
450297
55645
3452
2920
000
00000
00000
00000
00000
00000
00000
00000
0010
030928
238389
26192
185483
2000023
004
62000
00000
0000023
004
62000
00000
00
minus015
1119899293
2224
004
2902964
2001670
13088
01845
29345
01835
29345
000
00000
00025
1539138
7544
291
1039052
4970
186
20046
8529738
01876
10582
02690
19134
00544
09353
050
235214
2737579
16906
61920840
20000
4600923
00032
006
4100073
01284
000
00000
00075
56901
281822
45716
225576
20000
00000
00000
00000
00000
00000
00000
00000
0010
017992
51529
15163
36973
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899698
4825
00991
06240
2002436
15781
04239
404
2602224
15781
01927
15781
025
434800
664
668017
2750735
40004590
2001596
14118
00549
02968
02765
21869
00203
02249
050
107201
675811
80878
476895
20000
00000
00000
00000
0000058
01155
000
00000
00075
31529
163854
25990
110918
20000
00000
00000
00000
00000
00000
00000
00000
0010
015118
63845
13198
49453
20000
00000
0000016
00324
00016
00318
000
00000
00
24
minus005
1119899478
2082
01248
04836
2004707
39003
02017
20990
01368
18118
00959
18118
025
43205517
875044
4452338760
1060
43535
1304991
21418
06032
19032
046
0720265
02283
13231
050
11780220
66191783
13054869
65948027
1900758
10456
00313
02363
01334
10456
00161
02363
075
1216867
4261994
1562529
4708125
2000056
01126
00198
03359
00174
01871
000
00000
0010
0888475
5439067
1144057
69244
1720
000
00000
00000
00000
00000
00000
00000
00000
00
minus010
11198991074
12050
02496
26364
2007378
42545
03868
24023
02577
18340
00541
06057
025
24953166
5300
6630
28547095
6344
0156
1104324
22382
05843
29043
06883
25478
03452
22382
050
3116322
31612798
3661781
38159717
2000137
0118
400087
00883
00075
01028
000
00000
00075
513685
2435858
672926
2964331
20000
00000
0000015
00306
00112
01371
000
00000
0010
0270431
1462391
35260
91960620
20000
00000
0000052
01037
00039
00771
000
00000
00
minus015
1119899674
2613
01693
06084
2007280
78202
03115
27017
02512
34255
00284
05683
025
23417410
61951876
253876
736371360
415
09250
33429
02948
12902
02232
17855
00499
04354
050
4118014
36622393
4749373
39276216
2000114
0119
300053
01058
000
00000
00000
00000
00075
661765
5958864
874869
7920327
2000073
01458
00075
01509
000
4300867
000
00000
0010
0837932
11995859
968626
13244172
2000000
00000
00000
00000
00000
00000
00000
00000
minus020
11198993126
15554
07082
31980
2001942
32396
04290
33146
03196
32396
01669
32396
025
9333827
72575949
9993
219
80682031
1500589
02822
006
4607914
00844
07331
00204
02057
050
384242
2271989
508384
2808486
2000038
00760
00130
02596
00130
02596
00000
00000
075
570054
4053320
7017
325311990
20000
00000
00000
00000
00000
00000
00000
00000
0010
0233727
1202819
2915
7413260
0620
000
00000
00000
00000
00000
00000
00000
00000
00NoteldquoO
Frdquodeno
testhe
numbero
finstances
in20
setsof
datathatcanbe
solved
inlessthan
108no
desb
yusingtheb
ranch-
and-bo
undmetho
d
10 Mathematical Problems in Engineering
000
005
010
015
020
025
030
GA1 GA2 GA3 GA4GA
n16
n20n12
n24
Mea
n er
ror (
)
Figure 8 The performance of the genetic algorithms for various 119899
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
Mea
n er
ror (
)
minus005
minus010 minus020
minus015
Figure 9 The performance of the genetic algorithms for various 119886
percentage among all the 80 scenarios with varying learningrates numbers of jobs and values of 120579 The maximummean error percentages of GA
1 GA2 GA3 and GA
4were
16268 06032 08700 and 03452 respectively Themaximum mean error percentage of GA
1was more than
four times that of GA4 The combined algorithm GA
4also
clearly outperformed each of the three algorithms in terms
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
02505
0751
1n
Mea
n er
ror (
)
Figure 10The performance of the genetic algorithms for various 120579
of the maximum error percentage The worst cases of thecorresponding algorithms were 78202 69562 42205and 32396 respectively The worst case error of GA
1was
more than twice that of GA4Thus we would recommend the
combined algorithm GA4
6 Conclusions
In this paper we address a single-machine total completiontime problem with the sum of processing times basedlearning effect and release timesThe objective is to minimizethe total completion time The problem without learningconsideration is NP-hard one Thus we develop a branch-and-bound algorithm incorporatingwith several dominancesand two lower bounds to derive optimal solution and geneticalgorithms that was proposed to obtain near-optimal solu-tions respectively
The branch-and-bound algorithm performs well to solvethe instances of less than or equal to 24 jobs in a reasonableamount of time In the different varying parameter theproposed genetic algorithms are quite accurate for small sizeproblems and the mean error percentage of the proposedgenetic algorithms are less than 035 Another interestingtopic for future study is to consider and study the problem inthe multimachine environments or multicriteria cases
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 5
of chromosomes for the next generation This approach hasgained increasing popularity in solving many combinatorialoptimization problems in a wide variety of different disci-plines
431 Initial Settings In a GA every problem is presented bya code and each code is seen as a geneThe existing genes canbe combined and seen as a chromosome each of which is oneof the feasible solutions to a problem However traditionalrepresentation of GA does not work for scheduling problems(Etiler et al [46]) In dealing with this condition this studyadopts the same method that a structure can describe thejobs as a sequence in the problem To specify our approachseveral initial sequences are adopted In GA
1 jobs are placed
according to the shortest processing times (SPT) first ruleIn GA
2 jobs are arranged in earliest ready times (ERT) first
rule In GA3 jobs are arranged in a nondecreasing order on
the sum of job processing times and ready times Note thatbefore performing GA NEH algorithm (Nawaz et al [47]) isutilized to improve the quality of the solutions obtained fromthe previous rules to reduce many idle periods The processof GA
1 GA2 and GA
3are different initial sequences and use
the same selection crossover mutation operators populationsize and generations to obtain near-optimal solution Inaddition the fourth genetic algorithm denoted as GA
4 is
the best one among GA1 GA2 and GA
3 that is GA
4=
minGA1GA2GA3
In order to avoid rapidly observing a local optimum in asmall population or consume more waiting time in a largeone this study set a suitable population size as 60 (119873 =
60) in a preliminary trial It is also an important work toevaluate the fitness of selected chromosomes that each ofthe chromosomes is included or excluded from a feasiblesolution The main goal of this study is to minimize thetotal completion time Assume that 119878
119894(119905) is the 119894th string
in the 119894th generation and the total completion time of 119878119894(119905)
is sum119899
119895=1119862119895(119878119894(119905)) Then the fitness function of 119878
119894(119905) can be
represented as 119891(119878119894(119905)) Following are the calculations of the
strings in fitness function
119891 (119878119894(119905)) = max
1le119897le119873
119899
sum
119895=1
119862119895(119878119897(119905))
minus
119899
sum
119895=1
119862119895(119878119894(119905))
(12)
Moreover it is also crucial work to ensure that the probabilityof selection for a sequence with lower value of the objectivefunction is higher Thus the probability 119875(119878
119894(119905)) can be
written as follows
119875 (119878119894(119905)) =
119891 (119878119894(119905))
sum119899
119894=1119891 (119878119894(119905))
(13)
432 Operators There are a few operators that are used inthis study Following are the descriptions of those operatorscrossover mutation and selection
(a) CrossoverThis is an operator that exchanges some of thegenes of the selected parents with the main concept being
005 010 015 020 025 030000
001
001
002
002
003
Mea
n er
ror (
)
00040 00192 00108 00103 0020600171Pm
Figure 1 The performance of the genetic algorithms for various 119875119898
at (119899 119886 120579119873 119892) = (20 minus005 05 60 100119899)
that the descendant can inherit the advantages of its parentsThis study applied the linear order crossover operator (LOX)proposed by Falkenauer and Bouffouix [48] and is one of thebetter performers among the others (Etiler et al [46]) Theprobability of crossover is set to 1
(b) Mutation The main object of mutation is to achieve foran overall optimal solution and to avoid a locally optimalone In this study the mutation rates (119875
119898) are set at 010
based on our preliminary experiment as shown in Figure 1For (119899 119886 120579119873 119892) = (20 minus005 05 60 100119899) 100 sets of datawere randomly generated to evaluate the performance of theproposed algorithms with varying values of 119875
119898 The results
showed that the proposed algorithmshad the leastmean errorpercentage at 119875
119898= 010
(c) Selection This is a process that determines the proba-bility of each chromosome and is used to decide the betterchromosomes with the better fitness value The evolutionimplemented in our algorithm is based on the elitist list Wecopy the best offspring and use them to generate some ofthe next generation The rest of the offspring are generatedfrom the parent chromosomes by the roulette wheel selectionmethod which can maintain the variety of genes
433 Stopping Criteria In the preliminary experimentthe proposed GAs are terminated after 100 lowast 119899 genera-tions as shown in Figures 2 and 3 For (119899 119886 120579119873 119875
119898) =
(20 minus005 05 60 010) the above 100 sets of randomly gen-erated data were used to evaluate the performance of theproposed algorithms with varying values of 119892 The resultsshowed that the least mean error percentage of the proposedalgorithms would stabilize with reasonable CPU time rangeafter 119892 = 100119899
5 Computational Experiment
A computational experiment was conducted to evaluatethe efficiency of the branch-and-bound algorithm and
6 Mathematical Problems in Engineering
00738 00510 00315 00258 00040
000
001
002
003
004
005
006
007
008
Mea
n er
ror (
)
40n 60n 80n 100n20n
g
Figure 2 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
01666 03260 04981 06800 08382
000
010
020
030
040
050
060
070
080
090
CPU
tim
es (s
)
g
40n 60n 80n 100n20n
Figure 3 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
the accuracies of the genetic algorithmsThe algorithms werecoded in Fortran and run on Compaq Visual Fortran version66 on an Intel(R)Core(TM)2QuadCPU266GHzwith 4GBRAM on Windows Vista The experimental design followedReeves [49] design The job processing times were generatedfrom a uniform distribution over the integers between 1 and20 in every case while the release times were generated froma uniform distribution over the integers on (0 20119899120579) where 119899is the number of jobs Five different sets of problem instanceswere generated by giving 120579 the values 1119899 025 05 075and 1
For the branch-and-bound algorithm the average and themaximum numbers of nodes as well as the average and themaximum execution times (in seconds) were recorded Forthe three genetic algorithms the mean and the maximum
0
200
400
600
800
1000
1200
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005
minus010
minus015
minus020
120579
1n
Figure 4Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 12
error percentages were recorded where the error percentagewas calculated as
(GA119894minus TClowast)
TClowastlowast 100 (14)
where GA119894(119894 = 1 2 3 4) is the total completion time
obtained from the genetic algorithmand TClowast is the totalcompletion time of the optimal schedule The computationaltimes of the heuristic algorithmswere not recorded since theywere finished within a second
In the computational experiment four different numbersof jobs (119899 = 12 16 20 and 24) four different values of learn-ing effect (119886 = minus005 minus010 minus015 andminus020) and five differ-ent values of generation parameter of release times (120579 = 1119899025 05 075 and 1) were tested in the branch-and-boundalgorithmAs a consequence 80 experimental situationswereexamined A set of 20 instances were randomly generatedfor each situation and a total of 1600 problems were testedThe algorithms were set to skip to the next set of data if thenumber of nodes exceeded 108 The results are presented inTable 1 and Figures 4 5 6 and 7 Figures 4ndash7 showed theaverage number of nodes for various 120579 and 119886 at job size 1216 20 and 24 respectively The average number of nodesdecreased as the value of 120579 increased when 119899 was greaterthan 16 This was the direct result of the efficiency of LB
1
and LB2 As 120579 increased the frequency of applications of LB
2
would increase Consequently it would yield longer releasetimes in those cases and the properties were more powerfulMoreover LB
1is more efficient than LB
2 Table 1 and Figures
4ndash7 also showed whether in job size the algorithms had theleast mean number of nodes at 120579 = 1119899 It was due to thefact that with 120579 = 1119899 the release time was relatively shortand the completion time would readily exceed the release
Mathematical Problems in Engineering 7
0
10000
20000
30000
40000
50000
60000
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005minus010
minus015
minus020
120579
1n
Figure 5The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 16
time In those cases Property 6 was applied more frequentlyconversely the completion time would not easily exceed therelease time when the values of 120579 increased Moreover thenumber of nodes increased exponentially as the number ofjobs increased which was typical of an NP-hard problem Asillustrated in Table 1 when 119899 = 24 there were five cases inwhich the branch-and-bound algorithm could solve all theproblems optimally larger than 10
8 nodes The branch-and-bound algorithm had the worst performance when (119899 119886 120579) =
(24 minus005 025)with 87times107 nodes and 5234 secondsWith
120579 fixed at 1119899 the decrease of the completion time would berelatively small at the beginning when the learning effectwassmall (eg 119886 = minus005) In other words the completion timewould easily exceed the release time which would expeditethe timing of invoking Property 6 and consequently theaverage number of nodes would be smaller With 120579 = 025 as119899 increased the corresponding least average number of nodeswould occur at greater values of learning effect
The performance of the proposed GA algorithms out ofthe 80 evaluations and a total of 1600 problems was testedThe number of times that each of the objective functions ofthe GA
1 GA2 and GA
3algorithms had the smallest mean
error percentage was 45 41 and 49 respectively In additionin Table 1 and Figures 8 9 and 10 their performanceswere not affected with the learning rate the generationparameter 120579 of release times or the number of jobs Noneof the three genetic algorithms had absolutely dominantperformance in terms of mean error percentage Howeverthe combined algorithm GA
4strikingly outperformed each
of the three algorithms in terms of the maximummean error
0
500000
1000000
1500000
2000000
2500000
3000000
3500000
4000000
4500000
5000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 6Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 20
0
5000000
10000000
15000000
20000000
25000000
30000000
35000000
40000000
45000000
50000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 7The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 24
8 Mathematical Problems in Engineering
Table1Th
eperform
ance
oftheb
ranch-
and-bo
undandgenetic
algorithm
s
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
12
minus005
111989942
385
00031
00156
2000858
17151
000
00000
00004
0108013
000
00000
00025
549
1781
00187
00624
2001055
21091
00142
02839
000
00000
00000
00000
00050
473
1189
00140
00312
20000
00000
00000
00000
00000
00000
00000
00000
00075
414
1013
00125
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0569
2427
00164
00624
20000
00000
00000
00000
00000
00000
00000
00000
00
minus010
111989962
488
00023
00156
2001646
20568
01617
20568
000
00000
00000
00000
00025
1047
4318
00312
01092
20000
4400878
000
4400878
00283
05665
000
00000
00050
480
1321
00140
004
6820
000
00000
00006
8413
676
00019
00373
000
00000
00075
786
5291
00203
01404
20000
00000
00000
00000
00000
00000
00000
00000
0010
0396
1382
00125
004
6820
000
00000
00000
00000
00000
00000
00000
00000
00
minus015
111989921
7100016
00156
2000206
04130
00000
00000
00360
04130
00000
00000
025
817
4363
00226
00936
20000
00000
00000
00000
0001228
24567
000
00000
00050
493
1969
00148
00624
20000
00000
00000
00000
00000
00000
00000
00000
00075
286
803
000
9400312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0417
1259
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
111989947
336
00023
00156
2001295
25896
01295
25896
000
03000
69000
00000
00025
1091
6689
00289
01560
2000106
02119
000
00000
00000
00000
00000
00000
00050
337
948
00109
00312
2000314
06284
000
00000
00000
00000
00000
00000
00075
452
1121
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0319
1387
00086
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
16
minus005
111989974
365
00078
00312
2002066
34518
02202
27743
01114
12163
006
0812
163
025
39325
105847
20108
58188
2004522
53071
01479
17548
00621
04629
00309
03126
050
18779
107488
08697
44616
2000052
01035
00935
16983
00849
16983
000
00000
00075
22967
290574
09259
102025
20000
00000
00000
00000
00000
00000
00000
00000
0010
05144
34260
02535
15288
20000
4100811
00115
02306
000
00000
00000
00000
00
minus010
111989991
323
00086
00312
2001813
34958
02368
27957
01906
19030
000
00000
00025
53237
291654
25826
127297
2003151
25393
02824
21606
01918
21606
01271
21606
050
15022
193968
06778
81277
2000616
10305
00154
03088
00154
03088
00000
00000
075
15144
197196
06474
80029
20000
00000
00000
00000
00000
0600120
000
00000
0010
03403
9875
01794
04992
20000
00000
00000
00000
00000
00000
00000
00000
00
minus015
1119899118
1368
00101
01092
2002462
45183
00431
06619
00103
02068
000
00000
00025
52185
4846
6222994
190165
2000762
08319
03212
37507
00969
10191
00418
08319
050
17808
101503
08221
46020
2000740
07668
00356
06773
00574
06773
00339
06773
075
3763
16453
01997
08112
20000
00000
00000
00000
00000
6701343
000
00000
0010
03882
14411
01911
07020
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899200
828
00172
00624
2001832
26781
03397
1766
003644
22420
000
00000
00025
25881
149925
11224
53664
2000140
02772
00299
04614
000
9801922
000
0200034
050
11829
48284
05819
23556
20000
00000
0000072
01431
000
00000
00000
00000
00075
2687
12382
01427
05928
20000
00000
00000
00000
00000
00000
00000
00000
0010
02047
5258
01076
02340
20000
00000
00000
00000
00000
00000
00000
00000
00
Mathematical Problems in Engineering 9
Table1Con
tinued
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
20
minus005
111989982
278
00156
00624
2016
268
74028
03538
69562
00113
02265
000
00000
00025
1319111
4407360
1059941
3224697
2007803
29510
05928
32815
08700
42205
02480
15709
050
963361
9983475
618895
5687017
2000377
040
1601608
1746
402224
15708
00033
00510
075
323776
2518001
229485
1831606
2000022
004
46000
00000
0000027
004
46000
00000
0010
041188
138955
33642
1110
7220
00017
00338
00017
00338
000
00000
00000
00000
00
minus010
11198991345
14123
01716
17317
2006707
34958
04690
28787
04353
34958
00000
00000
025
2134707
16978059
1543318
12044214
2003845
23175
01848
06798
03536
21979
004
6002909
050
202434
1320363
156266
1024927
2000000
00000
000
4500786
00000
00000
00000
00000
075
67733
450297
55645
3452
2920
000
00000
00000
00000
00000
00000
00000
00000
0010
030928
238389
26192
185483
2000023
004
62000
00000
0000023
004
62000
00000
00
minus015
1119899293
2224
004
2902964
2001670
13088
01845
29345
01835
29345
000
00000
00025
1539138
7544
291
1039052
4970
186
20046
8529738
01876
10582
02690
19134
00544
09353
050
235214
2737579
16906
61920840
20000
4600923
00032
006
4100073
01284
000
00000
00075
56901
281822
45716
225576
20000
00000
00000
00000
00000
00000
00000
00000
0010
017992
51529
15163
36973
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899698
4825
00991
06240
2002436
15781
04239
404
2602224
15781
01927
15781
025
434800
664
668017
2750735
40004590
2001596
14118
00549
02968
02765
21869
00203
02249
050
107201
675811
80878
476895
20000
00000
00000
00000
0000058
01155
000
00000
00075
31529
163854
25990
110918
20000
00000
00000
00000
00000
00000
00000
00000
0010
015118
63845
13198
49453
20000
00000
0000016
00324
00016
00318
000
00000
00
24
minus005
1119899478
2082
01248
04836
2004707
39003
02017
20990
01368
18118
00959
18118
025
43205517
875044
4452338760
1060
43535
1304991
21418
06032
19032
046
0720265
02283
13231
050
11780220
66191783
13054869
65948027
1900758
10456
00313
02363
01334
10456
00161
02363
075
1216867
4261994
1562529
4708125
2000056
01126
00198
03359
00174
01871
000
00000
0010
0888475
5439067
1144057
69244
1720
000
00000
00000
00000
00000
00000
00000
00000
00
minus010
11198991074
12050
02496
26364
2007378
42545
03868
24023
02577
18340
00541
06057
025
24953166
5300
6630
28547095
6344
0156
1104324
22382
05843
29043
06883
25478
03452
22382
050
3116322
31612798
3661781
38159717
2000137
0118
400087
00883
00075
01028
000
00000
00075
513685
2435858
672926
2964331
20000
00000
0000015
00306
00112
01371
000
00000
0010
0270431
1462391
35260
91960620
20000
00000
0000052
01037
00039
00771
000
00000
00
minus015
1119899674
2613
01693
06084
2007280
78202
03115
27017
02512
34255
00284
05683
025
23417410
61951876
253876
736371360
415
09250
33429
02948
12902
02232
17855
00499
04354
050
4118014
36622393
4749373
39276216
2000114
0119
300053
01058
000
00000
00000
00000
00075
661765
5958864
874869
7920327
2000073
01458
00075
01509
000
4300867
000
00000
0010
0837932
11995859
968626
13244172
2000000
00000
00000
00000
00000
00000
00000
00000
minus020
11198993126
15554
07082
31980
2001942
32396
04290
33146
03196
32396
01669
32396
025
9333827
72575949
9993
219
80682031
1500589
02822
006
4607914
00844
07331
00204
02057
050
384242
2271989
508384
2808486
2000038
00760
00130
02596
00130
02596
00000
00000
075
570054
4053320
7017
325311990
20000
00000
00000
00000
00000
00000
00000
00000
0010
0233727
1202819
2915
7413260
0620
000
00000
00000
00000
00000
00000
00000
00000
00NoteldquoO
Frdquodeno
testhe
numbero
finstances
in20
setsof
datathatcanbe
solved
inlessthan
108no
desb
yusingtheb
ranch-
and-bo
undmetho
d
10 Mathematical Problems in Engineering
000
005
010
015
020
025
030
GA1 GA2 GA3 GA4GA
n16
n20n12
n24
Mea
n er
ror (
)
Figure 8 The performance of the genetic algorithms for various 119899
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
Mea
n er
ror (
)
minus005
minus010 minus020
minus015
Figure 9 The performance of the genetic algorithms for various 119886
percentage among all the 80 scenarios with varying learningrates numbers of jobs and values of 120579 The maximummean error percentages of GA
1 GA2 GA3 and GA
4were
16268 06032 08700 and 03452 respectively Themaximum mean error percentage of GA
1was more than
four times that of GA4 The combined algorithm GA
4also
clearly outperformed each of the three algorithms in terms
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
02505
0751
1n
Mea
n er
ror (
)
Figure 10The performance of the genetic algorithms for various 120579
of the maximum error percentage The worst cases of thecorresponding algorithms were 78202 69562 42205and 32396 respectively The worst case error of GA
1was
more than twice that of GA4Thus we would recommend the
combined algorithm GA4
6 Conclusions
In this paper we address a single-machine total completiontime problem with the sum of processing times basedlearning effect and release timesThe objective is to minimizethe total completion time The problem without learningconsideration is NP-hard one Thus we develop a branch-and-bound algorithm incorporatingwith several dominancesand two lower bounds to derive optimal solution and geneticalgorithms that was proposed to obtain near-optimal solu-tions respectively
The branch-and-bound algorithm performs well to solvethe instances of less than or equal to 24 jobs in a reasonableamount of time In the different varying parameter theproposed genetic algorithms are quite accurate for small sizeproblems and the mean error percentage of the proposedgenetic algorithms are less than 035 Another interestingtopic for future study is to consider and study the problem inthe multimachine environments or multicriteria cases
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
6 Mathematical Problems in Engineering
00738 00510 00315 00258 00040
000
001
002
003
004
005
006
007
008
Mea
n er
ror (
)
40n 60n 80n 100n20n
g
Figure 2 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
01666 03260 04981 06800 08382
000
010
020
030
040
050
060
070
080
090
CPU
tim
es (s
)
g
40n 60n 80n 100n20n
Figure 3 The performance of the genetic algorithms for various 119892at (119899 119886 120579119873 119875
119898) = (20 minus005 05 60 010)
the accuracies of the genetic algorithmsThe algorithms werecoded in Fortran and run on Compaq Visual Fortran version66 on an Intel(R)Core(TM)2QuadCPU266GHzwith 4GBRAM on Windows Vista The experimental design followedReeves [49] design The job processing times were generatedfrom a uniform distribution over the integers between 1 and20 in every case while the release times were generated froma uniform distribution over the integers on (0 20119899120579) where 119899is the number of jobs Five different sets of problem instanceswere generated by giving 120579 the values 1119899 025 05 075and 1
For the branch-and-bound algorithm the average and themaximum numbers of nodes as well as the average and themaximum execution times (in seconds) were recorded Forthe three genetic algorithms the mean and the maximum
0
200
400
600
800
1000
1200
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005
minus010
minus015
minus020
120579
1n
Figure 4Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 12
error percentages were recorded where the error percentagewas calculated as
(GA119894minus TClowast)
TClowastlowast 100 (14)
where GA119894(119894 = 1 2 3 4) is the total completion time
obtained from the genetic algorithmand TClowast is the totalcompletion time of the optimal schedule The computationaltimes of the heuristic algorithmswere not recorded since theywere finished within a second
In the computational experiment four different numbersof jobs (119899 = 12 16 20 and 24) four different values of learn-ing effect (119886 = minus005 minus010 minus015 andminus020) and five differ-ent values of generation parameter of release times (120579 = 1119899025 05 075 and 1) were tested in the branch-and-boundalgorithmAs a consequence 80 experimental situationswereexamined A set of 20 instances were randomly generatedfor each situation and a total of 1600 problems were testedThe algorithms were set to skip to the next set of data if thenumber of nodes exceeded 108 The results are presented inTable 1 and Figures 4 5 6 and 7 Figures 4ndash7 showed theaverage number of nodes for various 120579 and 119886 at job size 1216 20 and 24 respectively The average number of nodesdecreased as the value of 120579 increased when 119899 was greaterthan 16 This was the direct result of the efficiency of LB
1
and LB2 As 120579 increased the frequency of applications of LB
2
would increase Consequently it would yield longer releasetimes in those cases and the properties were more powerfulMoreover LB
1is more efficient than LB
2 Table 1 and Figures
4ndash7 also showed whether in job size the algorithms had theleast mean number of nodes at 120579 = 1119899 It was due to thefact that with 120579 = 1119899 the release time was relatively shortand the completion time would readily exceed the release
Mathematical Problems in Engineering 7
0
10000
20000
30000
40000
50000
60000
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005minus010
minus015
minus020
120579
1n
Figure 5The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 16
time In those cases Property 6 was applied more frequentlyconversely the completion time would not easily exceed therelease time when the values of 120579 increased Moreover thenumber of nodes increased exponentially as the number ofjobs increased which was typical of an NP-hard problem Asillustrated in Table 1 when 119899 = 24 there were five cases inwhich the branch-and-bound algorithm could solve all theproblems optimally larger than 10
8 nodes The branch-and-bound algorithm had the worst performance when (119899 119886 120579) =
(24 minus005 025)with 87times107 nodes and 5234 secondsWith
120579 fixed at 1119899 the decrease of the completion time would berelatively small at the beginning when the learning effectwassmall (eg 119886 = minus005) In other words the completion timewould easily exceed the release time which would expeditethe timing of invoking Property 6 and consequently theaverage number of nodes would be smaller With 120579 = 025 as119899 increased the corresponding least average number of nodeswould occur at greater values of learning effect
The performance of the proposed GA algorithms out ofthe 80 evaluations and a total of 1600 problems was testedThe number of times that each of the objective functions ofthe GA
1 GA2 and GA
3algorithms had the smallest mean
error percentage was 45 41 and 49 respectively In additionin Table 1 and Figures 8 9 and 10 their performanceswere not affected with the learning rate the generationparameter 120579 of release times or the number of jobs Noneof the three genetic algorithms had absolutely dominantperformance in terms of mean error percentage Howeverthe combined algorithm GA
4strikingly outperformed each
of the three algorithms in terms of the maximummean error
0
500000
1000000
1500000
2000000
2500000
3000000
3500000
4000000
4500000
5000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 6Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 20
0
5000000
10000000
15000000
20000000
25000000
30000000
35000000
40000000
45000000
50000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 7The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 24
8 Mathematical Problems in Engineering
Table1Th
eperform
ance
oftheb
ranch-
and-bo
undandgenetic
algorithm
s
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
12
minus005
111989942
385
00031
00156
2000858
17151
000
00000
00004
0108013
000
00000
00025
549
1781
00187
00624
2001055
21091
00142
02839
000
00000
00000
00000
00050
473
1189
00140
00312
20000
00000
00000
00000
00000
00000
00000
00000
00075
414
1013
00125
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0569
2427
00164
00624
20000
00000
00000
00000
00000
00000
00000
00000
00
minus010
111989962
488
00023
00156
2001646
20568
01617
20568
000
00000
00000
00000
00025
1047
4318
00312
01092
20000
4400878
000
4400878
00283
05665
000
00000
00050
480
1321
00140
004
6820
000
00000
00006
8413
676
00019
00373
000
00000
00075
786
5291
00203
01404
20000
00000
00000
00000
00000
00000
00000
00000
0010
0396
1382
00125
004
6820
000
00000
00000
00000
00000
00000
00000
00000
00
minus015
111989921
7100016
00156
2000206
04130
00000
00000
00360
04130
00000
00000
025
817
4363
00226
00936
20000
00000
00000
00000
0001228
24567
000
00000
00050
493
1969
00148
00624
20000
00000
00000
00000
00000
00000
00000
00000
00075
286
803
000
9400312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0417
1259
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
111989947
336
00023
00156
2001295
25896
01295
25896
000
03000
69000
00000
00025
1091
6689
00289
01560
2000106
02119
000
00000
00000
00000
00000
00000
00050
337
948
00109
00312
2000314
06284
000
00000
00000
00000
00000
00000
00075
452
1121
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0319
1387
00086
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
16
minus005
111989974
365
00078
00312
2002066
34518
02202
27743
01114
12163
006
0812
163
025
39325
105847
20108
58188
2004522
53071
01479
17548
00621
04629
00309
03126
050
18779
107488
08697
44616
2000052
01035
00935
16983
00849
16983
000
00000
00075
22967
290574
09259
102025
20000
00000
00000
00000
00000
00000
00000
00000
0010
05144
34260
02535
15288
20000
4100811
00115
02306
000
00000
00000
00000
00
minus010
111989991
323
00086
00312
2001813
34958
02368
27957
01906
19030
000
00000
00025
53237
291654
25826
127297
2003151
25393
02824
21606
01918
21606
01271
21606
050
15022
193968
06778
81277
2000616
10305
00154
03088
00154
03088
00000
00000
075
15144
197196
06474
80029
20000
00000
00000
00000
00000
0600120
000
00000
0010
03403
9875
01794
04992
20000
00000
00000
00000
00000
00000
00000
00000
00
minus015
1119899118
1368
00101
01092
2002462
45183
00431
06619
00103
02068
000
00000
00025
52185
4846
6222994
190165
2000762
08319
03212
37507
00969
10191
00418
08319
050
17808
101503
08221
46020
2000740
07668
00356
06773
00574
06773
00339
06773
075
3763
16453
01997
08112
20000
00000
00000
00000
00000
6701343
000
00000
0010
03882
14411
01911
07020
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899200
828
00172
00624
2001832
26781
03397
1766
003644
22420
000
00000
00025
25881
149925
11224
53664
2000140
02772
00299
04614
000
9801922
000
0200034
050
11829
48284
05819
23556
20000
00000
0000072
01431
000
00000
00000
00000
00075
2687
12382
01427
05928
20000
00000
00000
00000
00000
00000
00000
00000
0010
02047
5258
01076
02340
20000
00000
00000
00000
00000
00000
00000
00000
00
Mathematical Problems in Engineering 9
Table1Con
tinued
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
20
minus005
111989982
278
00156
00624
2016
268
74028
03538
69562
00113
02265
000
00000
00025
1319111
4407360
1059941
3224697
2007803
29510
05928
32815
08700
42205
02480
15709
050
963361
9983475
618895
5687017
2000377
040
1601608
1746
402224
15708
00033
00510
075
323776
2518001
229485
1831606
2000022
004
46000
00000
0000027
004
46000
00000
0010
041188
138955
33642
1110
7220
00017
00338
00017
00338
000
00000
00000
00000
00
minus010
11198991345
14123
01716
17317
2006707
34958
04690
28787
04353
34958
00000
00000
025
2134707
16978059
1543318
12044214
2003845
23175
01848
06798
03536
21979
004
6002909
050
202434
1320363
156266
1024927
2000000
00000
000
4500786
00000
00000
00000
00000
075
67733
450297
55645
3452
2920
000
00000
00000
00000
00000
00000
00000
00000
0010
030928
238389
26192
185483
2000023
004
62000
00000
0000023
004
62000
00000
00
minus015
1119899293
2224
004
2902964
2001670
13088
01845
29345
01835
29345
000
00000
00025
1539138
7544
291
1039052
4970
186
20046
8529738
01876
10582
02690
19134
00544
09353
050
235214
2737579
16906
61920840
20000
4600923
00032
006
4100073
01284
000
00000
00075
56901
281822
45716
225576
20000
00000
00000
00000
00000
00000
00000
00000
0010
017992
51529
15163
36973
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899698
4825
00991
06240
2002436
15781
04239
404
2602224
15781
01927
15781
025
434800
664
668017
2750735
40004590
2001596
14118
00549
02968
02765
21869
00203
02249
050
107201
675811
80878
476895
20000
00000
00000
00000
0000058
01155
000
00000
00075
31529
163854
25990
110918
20000
00000
00000
00000
00000
00000
00000
00000
0010
015118
63845
13198
49453
20000
00000
0000016
00324
00016
00318
000
00000
00
24
minus005
1119899478
2082
01248
04836
2004707
39003
02017
20990
01368
18118
00959
18118
025
43205517
875044
4452338760
1060
43535
1304991
21418
06032
19032
046
0720265
02283
13231
050
11780220
66191783
13054869
65948027
1900758
10456
00313
02363
01334
10456
00161
02363
075
1216867
4261994
1562529
4708125
2000056
01126
00198
03359
00174
01871
000
00000
0010
0888475
5439067
1144057
69244
1720
000
00000
00000
00000
00000
00000
00000
00000
00
minus010
11198991074
12050
02496
26364
2007378
42545
03868
24023
02577
18340
00541
06057
025
24953166
5300
6630
28547095
6344
0156
1104324
22382
05843
29043
06883
25478
03452
22382
050
3116322
31612798
3661781
38159717
2000137
0118
400087
00883
00075
01028
000
00000
00075
513685
2435858
672926
2964331
20000
00000
0000015
00306
00112
01371
000
00000
0010
0270431
1462391
35260
91960620
20000
00000
0000052
01037
00039
00771
000
00000
00
minus015
1119899674
2613
01693
06084
2007280
78202
03115
27017
02512
34255
00284
05683
025
23417410
61951876
253876
736371360
415
09250
33429
02948
12902
02232
17855
00499
04354
050
4118014
36622393
4749373
39276216
2000114
0119
300053
01058
000
00000
00000
00000
00075
661765
5958864
874869
7920327
2000073
01458
00075
01509
000
4300867
000
00000
0010
0837932
11995859
968626
13244172
2000000
00000
00000
00000
00000
00000
00000
00000
minus020
11198993126
15554
07082
31980
2001942
32396
04290
33146
03196
32396
01669
32396
025
9333827
72575949
9993
219
80682031
1500589
02822
006
4607914
00844
07331
00204
02057
050
384242
2271989
508384
2808486
2000038
00760
00130
02596
00130
02596
00000
00000
075
570054
4053320
7017
325311990
20000
00000
00000
00000
00000
00000
00000
00000
0010
0233727
1202819
2915
7413260
0620
000
00000
00000
00000
00000
00000
00000
00000
00NoteldquoO
Frdquodeno
testhe
numbero
finstances
in20
setsof
datathatcanbe
solved
inlessthan
108no
desb
yusingtheb
ranch-
and-bo
undmetho
d
10 Mathematical Problems in Engineering
000
005
010
015
020
025
030
GA1 GA2 GA3 GA4GA
n16
n20n12
n24
Mea
n er
ror (
)
Figure 8 The performance of the genetic algorithms for various 119899
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
Mea
n er
ror (
)
minus005
minus010 minus020
minus015
Figure 9 The performance of the genetic algorithms for various 119886
percentage among all the 80 scenarios with varying learningrates numbers of jobs and values of 120579 The maximummean error percentages of GA
1 GA2 GA3 and GA
4were
16268 06032 08700 and 03452 respectively Themaximum mean error percentage of GA
1was more than
four times that of GA4 The combined algorithm GA
4also
clearly outperformed each of the three algorithms in terms
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
02505
0751
1n
Mea
n er
ror (
)
Figure 10The performance of the genetic algorithms for various 120579
of the maximum error percentage The worst cases of thecorresponding algorithms were 78202 69562 42205and 32396 respectively The worst case error of GA
1was
more than twice that of GA4Thus we would recommend the
combined algorithm GA4
6 Conclusions
In this paper we address a single-machine total completiontime problem with the sum of processing times basedlearning effect and release timesThe objective is to minimizethe total completion time The problem without learningconsideration is NP-hard one Thus we develop a branch-and-bound algorithm incorporatingwith several dominancesand two lower bounds to derive optimal solution and geneticalgorithms that was proposed to obtain near-optimal solu-tions respectively
The branch-and-bound algorithm performs well to solvethe instances of less than or equal to 24 jobs in a reasonableamount of time In the different varying parameter theproposed genetic algorithms are quite accurate for small sizeproblems and the mean error percentage of the proposedgenetic algorithms are less than 035 Another interestingtopic for future study is to consider and study the problem inthe multimachine environments or multicriteria cases
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 7
0
10000
20000
30000
40000
50000
60000
025 05 075 1
Aver
age n
umbe
r of n
odes
minus005minus010
minus015
minus020
120579
1n
Figure 5The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 16
time In those cases Property 6 was applied more frequentlyconversely the completion time would not easily exceed therelease time when the values of 120579 increased Moreover thenumber of nodes increased exponentially as the number ofjobs increased which was typical of an NP-hard problem Asillustrated in Table 1 when 119899 = 24 there were five cases inwhich the branch-and-bound algorithm could solve all theproblems optimally larger than 10
8 nodes The branch-and-bound algorithm had the worst performance when (119899 119886 120579) =
(24 minus005 025)with 87times107 nodes and 5234 secondsWith
120579 fixed at 1119899 the decrease of the completion time would berelatively small at the beginning when the learning effectwassmall (eg 119886 = minus005) In other words the completion timewould easily exceed the release time which would expeditethe timing of invoking Property 6 and consequently theaverage number of nodes would be smaller With 120579 = 025 as119899 increased the corresponding least average number of nodeswould occur at greater values of learning effect
The performance of the proposed GA algorithms out ofthe 80 evaluations and a total of 1600 problems was testedThe number of times that each of the objective functions ofthe GA
1 GA2 and GA
3algorithms had the smallest mean
error percentage was 45 41 and 49 respectively In additionin Table 1 and Figures 8 9 and 10 their performanceswere not affected with the learning rate the generationparameter 120579 of release times or the number of jobs Noneof the three genetic algorithms had absolutely dominantperformance in terms of mean error percentage Howeverthe combined algorithm GA
4strikingly outperformed each
of the three algorithms in terms of the maximummean error
0
500000
1000000
1500000
2000000
2500000
3000000
3500000
4000000
4500000
5000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 6Theperformance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 20
0
5000000
10000000
15000000
20000000
25000000
30000000
35000000
40000000
45000000
50000000
025 05 075 1
Aver
age n
umbe
r of n
odes
120579
1n
minus005minus010
minus015
minus020
Figure 7The performance of the branch-and-bound algorithms forvarious 120579 and 119886 at 119899 = 24
8 Mathematical Problems in Engineering
Table1Th
eperform
ance
oftheb
ranch-
and-bo
undandgenetic
algorithm
s
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
12
minus005
111989942
385
00031
00156
2000858
17151
000
00000
00004
0108013
000
00000
00025
549
1781
00187
00624
2001055
21091
00142
02839
000
00000
00000
00000
00050
473
1189
00140
00312
20000
00000
00000
00000
00000
00000
00000
00000
00075
414
1013
00125
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0569
2427
00164
00624
20000
00000
00000
00000
00000
00000
00000
00000
00
minus010
111989962
488
00023
00156
2001646
20568
01617
20568
000
00000
00000
00000
00025
1047
4318
00312
01092
20000
4400878
000
4400878
00283
05665
000
00000
00050
480
1321
00140
004
6820
000
00000
00006
8413
676
00019
00373
000
00000
00075
786
5291
00203
01404
20000
00000
00000
00000
00000
00000
00000
00000
0010
0396
1382
00125
004
6820
000
00000
00000
00000
00000
00000
00000
00000
00
minus015
111989921
7100016
00156
2000206
04130
00000
00000
00360
04130
00000
00000
025
817
4363
00226
00936
20000
00000
00000
00000
0001228
24567
000
00000
00050
493
1969
00148
00624
20000
00000
00000
00000
00000
00000
00000
00000
00075
286
803
000
9400312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0417
1259
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
111989947
336
00023
00156
2001295
25896
01295
25896
000
03000
69000
00000
00025
1091
6689
00289
01560
2000106
02119
000
00000
00000
00000
00000
00000
00050
337
948
00109
00312
2000314
06284
000
00000
00000
00000
00000
00000
00075
452
1121
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0319
1387
00086
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
16
minus005
111989974
365
00078
00312
2002066
34518
02202
27743
01114
12163
006
0812
163
025
39325
105847
20108
58188
2004522
53071
01479
17548
00621
04629
00309
03126
050
18779
107488
08697
44616
2000052
01035
00935
16983
00849
16983
000
00000
00075
22967
290574
09259
102025
20000
00000
00000
00000
00000
00000
00000
00000
0010
05144
34260
02535
15288
20000
4100811
00115
02306
000
00000
00000
00000
00
minus010
111989991
323
00086
00312
2001813
34958
02368
27957
01906
19030
000
00000
00025
53237
291654
25826
127297
2003151
25393
02824
21606
01918
21606
01271
21606
050
15022
193968
06778
81277
2000616
10305
00154
03088
00154
03088
00000
00000
075
15144
197196
06474
80029
20000
00000
00000
00000
00000
0600120
000
00000
0010
03403
9875
01794
04992
20000
00000
00000
00000
00000
00000
00000
00000
00
minus015
1119899118
1368
00101
01092
2002462
45183
00431
06619
00103
02068
000
00000
00025
52185
4846
6222994
190165
2000762
08319
03212
37507
00969
10191
00418
08319
050
17808
101503
08221
46020
2000740
07668
00356
06773
00574
06773
00339
06773
075
3763
16453
01997
08112
20000
00000
00000
00000
00000
6701343
000
00000
0010
03882
14411
01911
07020
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899200
828
00172
00624
2001832
26781
03397
1766
003644
22420
000
00000
00025
25881
149925
11224
53664
2000140
02772
00299
04614
000
9801922
000
0200034
050
11829
48284
05819
23556
20000
00000
0000072
01431
000
00000
00000
00000
00075
2687
12382
01427
05928
20000
00000
00000
00000
00000
00000
00000
00000
0010
02047
5258
01076
02340
20000
00000
00000
00000
00000
00000
00000
00000
00
Mathematical Problems in Engineering 9
Table1Con
tinued
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
20
minus005
111989982
278
00156
00624
2016
268
74028
03538
69562
00113
02265
000
00000
00025
1319111
4407360
1059941
3224697
2007803
29510
05928
32815
08700
42205
02480
15709
050
963361
9983475
618895
5687017
2000377
040
1601608
1746
402224
15708
00033
00510
075
323776
2518001
229485
1831606
2000022
004
46000
00000
0000027
004
46000
00000
0010
041188
138955
33642
1110
7220
00017
00338
00017
00338
000
00000
00000
00000
00
minus010
11198991345
14123
01716
17317
2006707
34958
04690
28787
04353
34958
00000
00000
025
2134707
16978059
1543318
12044214
2003845
23175
01848
06798
03536
21979
004
6002909
050
202434
1320363
156266
1024927
2000000
00000
000
4500786
00000
00000
00000
00000
075
67733
450297
55645
3452
2920
000
00000
00000
00000
00000
00000
00000
00000
0010
030928
238389
26192
185483
2000023
004
62000
00000
0000023
004
62000
00000
00
minus015
1119899293
2224
004
2902964
2001670
13088
01845
29345
01835
29345
000
00000
00025
1539138
7544
291
1039052
4970
186
20046
8529738
01876
10582
02690
19134
00544
09353
050
235214
2737579
16906
61920840
20000
4600923
00032
006
4100073
01284
000
00000
00075
56901
281822
45716
225576
20000
00000
00000
00000
00000
00000
00000
00000
0010
017992
51529
15163
36973
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899698
4825
00991
06240
2002436
15781
04239
404
2602224
15781
01927
15781
025
434800
664
668017
2750735
40004590
2001596
14118
00549
02968
02765
21869
00203
02249
050
107201
675811
80878
476895
20000
00000
00000
00000
0000058
01155
000
00000
00075
31529
163854
25990
110918
20000
00000
00000
00000
00000
00000
00000
00000
0010
015118
63845
13198
49453
20000
00000
0000016
00324
00016
00318
000
00000
00
24
minus005
1119899478
2082
01248
04836
2004707
39003
02017
20990
01368
18118
00959
18118
025
43205517
875044
4452338760
1060
43535
1304991
21418
06032
19032
046
0720265
02283
13231
050
11780220
66191783
13054869
65948027
1900758
10456
00313
02363
01334
10456
00161
02363
075
1216867
4261994
1562529
4708125
2000056
01126
00198
03359
00174
01871
000
00000
0010
0888475
5439067
1144057
69244
1720
000
00000
00000
00000
00000
00000
00000
00000
00
minus010
11198991074
12050
02496
26364
2007378
42545
03868
24023
02577
18340
00541
06057
025
24953166
5300
6630
28547095
6344
0156
1104324
22382
05843
29043
06883
25478
03452
22382
050
3116322
31612798
3661781
38159717
2000137
0118
400087
00883
00075
01028
000
00000
00075
513685
2435858
672926
2964331
20000
00000
0000015
00306
00112
01371
000
00000
0010
0270431
1462391
35260
91960620
20000
00000
0000052
01037
00039
00771
000
00000
00
minus015
1119899674
2613
01693
06084
2007280
78202
03115
27017
02512
34255
00284
05683
025
23417410
61951876
253876
736371360
415
09250
33429
02948
12902
02232
17855
00499
04354
050
4118014
36622393
4749373
39276216
2000114
0119
300053
01058
000
00000
00000
00000
00075
661765
5958864
874869
7920327
2000073
01458
00075
01509
000
4300867
000
00000
0010
0837932
11995859
968626
13244172
2000000
00000
00000
00000
00000
00000
00000
00000
minus020
11198993126
15554
07082
31980
2001942
32396
04290
33146
03196
32396
01669
32396
025
9333827
72575949
9993
219
80682031
1500589
02822
006
4607914
00844
07331
00204
02057
050
384242
2271989
508384
2808486
2000038
00760
00130
02596
00130
02596
00000
00000
075
570054
4053320
7017
325311990
20000
00000
00000
00000
00000
00000
00000
00000
0010
0233727
1202819
2915
7413260
0620
000
00000
00000
00000
00000
00000
00000
00000
00NoteldquoO
Frdquodeno
testhe
numbero
finstances
in20
setsof
datathatcanbe
solved
inlessthan
108no
desb
yusingtheb
ranch-
and-bo
undmetho
d
10 Mathematical Problems in Engineering
000
005
010
015
020
025
030
GA1 GA2 GA3 GA4GA
n16
n20n12
n24
Mea
n er
ror (
)
Figure 8 The performance of the genetic algorithms for various 119899
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
Mea
n er
ror (
)
minus005
minus010 minus020
minus015
Figure 9 The performance of the genetic algorithms for various 119886
percentage among all the 80 scenarios with varying learningrates numbers of jobs and values of 120579 The maximummean error percentages of GA
1 GA2 GA3 and GA
4were
16268 06032 08700 and 03452 respectively Themaximum mean error percentage of GA
1was more than
four times that of GA4 The combined algorithm GA
4also
clearly outperformed each of the three algorithms in terms
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
02505
0751
1n
Mea
n er
ror (
)
Figure 10The performance of the genetic algorithms for various 120579
of the maximum error percentage The worst cases of thecorresponding algorithms were 78202 69562 42205and 32396 respectively The worst case error of GA
1was
more than twice that of GA4Thus we would recommend the
combined algorithm GA4
6 Conclusions
In this paper we address a single-machine total completiontime problem with the sum of processing times basedlearning effect and release timesThe objective is to minimizethe total completion time The problem without learningconsideration is NP-hard one Thus we develop a branch-and-bound algorithm incorporatingwith several dominancesand two lower bounds to derive optimal solution and geneticalgorithms that was proposed to obtain near-optimal solu-tions respectively
The branch-and-bound algorithm performs well to solvethe instances of less than or equal to 24 jobs in a reasonableamount of time In the different varying parameter theproposed genetic algorithms are quite accurate for small sizeproblems and the mean error percentage of the proposedgenetic algorithms are less than 035 Another interestingtopic for future study is to consider and study the problem inthe multimachine environments or multicriteria cases
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
8 Mathematical Problems in Engineering
Table1Th
eperform
ance
oftheb
ranch-
and-bo
undandgenetic
algorithm
s
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
12
minus005
111989942
385
00031
00156
2000858
17151
000
00000
00004
0108013
000
00000
00025
549
1781
00187
00624
2001055
21091
00142
02839
000
00000
00000
00000
00050
473
1189
00140
00312
20000
00000
00000
00000
00000
00000
00000
00000
00075
414
1013
00125
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0569
2427
00164
00624
20000
00000
00000
00000
00000
00000
00000
00000
00
minus010
111989962
488
00023
00156
2001646
20568
01617
20568
000
00000
00000
00000
00025
1047
4318
00312
01092
20000
4400878
000
4400878
00283
05665
000
00000
00050
480
1321
00140
004
6820
000
00000
00006
8413
676
00019
00373
000
00000
00075
786
5291
00203
01404
20000
00000
00000
00000
00000
00000
00000
00000
0010
0396
1382
00125
004
6820
000
00000
00000
00000
00000
00000
00000
00000
00
minus015
111989921
7100016
00156
2000206
04130
00000
00000
00360
04130
00000
00000
025
817
4363
00226
00936
20000
00000
00000
00000
0001228
24567
000
00000
00050
493
1969
00148
00624
20000
00000
00000
00000
00000
00000
00000
00000
00075
286
803
000
9400312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0417
1259
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
111989947
336
00023
00156
2001295
25896
01295
25896
000
03000
69000
00000
00025
1091
6689
00289
01560
2000106
02119
000
00000
00000
00000
00000
00000
00050
337
948
00109
00312
2000314
06284
000
00000
00000
00000
00000
00000
00075
452
1121
00117
00312
20000
00000
00000
00000
00000
00000
00000
00000
0010
0319
1387
00086
00312
20000
00000
00000
00000
00000
00000
00000
00000
00
16
minus005
111989974
365
00078
00312
2002066
34518
02202
27743
01114
12163
006
0812
163
025
39325
105847
20108
58188
2004522
53071
01479
17548
00621
04629
00309
03126
050
18779
107488
08697
44616
2000052
01035
00935
16983
00849
16983
000
00000
00075
22967
290574
09259
102025
20000
00000
00000
00000
00000
00000
00000
00000
0010
05144
34260
02535
15288
20000
4100811
00115
02306
000
00000
00000
00000
00
minus010
111989991
323
00086
00312
2001813
34958
02368
27957
01906
19030
000
00000
00025
53237
291654
25826
127297
2003151
25393
02824
21606
01918
21606
01271
21606
050
15022
193968
06778
81277
2000616
10305
00154
03088
00154
03088
00000
00000
075
15144
197196
06474
80029
20000
00000
00000
00000
00000
0600120
000
00000
0010
03403
9875
01794
04992
20000
00000
00000
00000
00000
00000
00000
00000
00
minus015
1119899118
1368
00101
01092
2002462
45183
00431
06619
00103
02068
000
00000
00025
52185
4846
6222994
190165
2000762
08319
03212
37507
00969
10191
00418
08319
050
17808
101503
08221
46020
2000740
07668
00356
06773
00574
06773
00339
06773
075
3763
16453
01997
08112
20000
00000
00000
00000
00000
6701343
000
00000
0010
03882
14411
01911
07020
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899200
828
00172
00624
2001832
26781
03397
1766
003644
22420
000
00000
00025
25881
149925
11224
53664
2000140
02772
00299
04614
000
9801922
000
0200034
050
11829
48284
05819
23556
20000
00000
0000072
01431
000
00000
00000
00000
00075
2687
12382
01427
05928
20000
00000
00000
00000
00000
00000
00000
00000
0010
02047
5258
01076
02340
20000
00000
00000
00000
00000
00000
00000
00000
00
Mathematical Problems in Engineering 9
Table1Con
tinued
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
20
minus005
111989982
278
00156
00624
2016
268
74028
03538
69562
00113
02265
000
00000
00025
1319111
4407360
1059941
3224697
2007803
29510
05928
32815
08700
42205
02480
15709
050
963361
9983475
618895
5687017
2000377
040
1601608
1746
402224
15708
00033
00510
075
323776
2518001
229485
1831606
2000022
004
46000
00000
0000027
004
46000
00000
0010
041188
138955
33642
1110
7220
00017
00338
00017
00338
000
00000
00000
00000
00
minus010
11198991345
14123
01716
17317
2006707
34958
04690
28787
04353
34958
00000
00000
025
2134707
16978059
1543318
12044214
2003845
23175
01848
06798
03536
21979
004
6002909
050
202434
1320363
156266
1024927
2000000
00000
000
4500786
00000
00000
00000
00000
075
67733
450297
55645
3452
2920
000
00000
00000
00000
00000
00000
00000
00000
0010
030928
238389
26192
185483
2000023
004
62000
00000
0000023
004
62000
00000
00
minus015
1119899293
2224
004
2902964
2001670
13088
01845
29345
01835
29345
000
00000
00025
1539138
7544
291
1039052
4970
186
20046
8529738
01876
10582
02690
19134
00544
09353
050
235214
2737579
16906
61920840
20000
4600923
00032
006
4100073
01284
000
00000
00075
56901
281822
45716
225576
20000
00000
00000
00000
00000
00000
00000
00000
0010
017992
51529
15163
36973
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899698
4825
00991
06240
2002436
15781
04239
404
2602224
15781
01927
15781
025
434800
664
668017
2750735
40004590
2001596
14118
00549
02968
02765
21869
00203
02249
050
107201
675811
80878
476895
20000
00000
00000
00000
0000058
01155
000
00000
00075
31529
163854
25990
110918
20000
00000
00000
00000
00000
00000
00000
00000
0010
015118
63845
13198
49453
20000
00000
0000016
00324
00016
00318
000
00000
00
24
minus005
1119899478
2082
01248
04836
2004707
39003
02017
20990
01368
18118
00959
18118
025
43205517
875044
4452338760
1060
43535
1304991
21418
06032
19032
046
0720265
02283
13231
050
11780220
66191783
13054869
65948027
1900758
10456
00313
02363
01334
10456
00161
02363
075
1216867
4261994
1562529
4708125
2000056
01126
00198
03359
00174
01871
000
00000
0010
0888475
5439067
1144057
69244
1720
000
00000
00000
00000
00000
00000
00000
00000
00
minus010
11198991074
12050
02496
26364
2007378
42545
03868
24023
02577
18340
00541
06057
025
24953166
5300
6630
28547095
6344
0156
1104324
22382
05843
29043
06883
25478
03452
22382
050
3116322
31612798
3661781
38159717
2000137
0118
400087
00883
00075
01028
000
00000
00075
513685
2435858
672926
2964331
20000
00000
0000015
00306
00112
01371
000
00000
0010
0270431
1462391
35260
91960620
20000
00000
0000052
01037
00039
00771
000
00000
00
minus015
1119899674
2613
01693
06084
2007280
78202
03115
27017
02512
34255
00284
05683
025
23417410
61951876
253876
736371360
415
09250
33429
02948
12902
02232
17855
00499
04354
050
4118014
36622393
4749373
39276216
2000114
0119
300053
01058
000
00000
00000
00000
00075
661765
5958864
874869
7920327
2000073
01458
00075
01509
000
4300867
000
00000
0010
0837932
11995859
968626
13244172
2000000
00000
00000
00000
00000
00000
00000
00000
minus020
11198993126
15554
07082
31980
2001942
32396
04290
33146
03196
32396
01669
32396
025
9333827
72575949
9993
219
80682031
1500589
02822
006
4607914
00844
07331
00204
02057
050
384242
2271989
508384
2808486
2000038
00760
00130
02596
00130
02596
00000
00000
075
570054
4053320
7017
325311990
20000
00000
00000
00000
00000
00000
00000
00000
0010
0233727
1202819
2915
7413260
0620
000
00000
00000
00000
00000
00000
00000
00000
00NoteldquoO
Frdquodeno
testhe
numbero
finstances
in20
setsof
datathatcanbe
solved
inlessthan
108no
desb
yusingtheb
ranch-
and-bo
undmetho
d
10 Mathematical Problems in Engineering
000
005
010
015
020
025
030
GA1 GA2 GA3 GA4GA
n16
n20n12
n24
Mea
n er
ror (
)
Figure 8 The performance of the genetic algorithms for various 119899
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
Mea
n er
ror (
)
minus005
minus010 minus020
minus015
Figure 9 The performance of the genetic algorithms for various 119886
percentage among all the 80 scenarios with varying learningrates numbers of jobs and values of 120579 The maximummean error percentages of GA
1 GA2 GA3 and GA
4were
16268 06032 08700 and 03452 respectively Themaximum mean error percentage of GA
1was more than
four times that of GA4 The combined algorithm GA
4also
clearly outperformed each of the three algorithms in terms
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
02505
0751
1n
Mea
n er
ror (
)
Figure 10The performance of the genetic algorithms for various 120579
of the maximum error percentage The worst cases of thecorresponding algorithms were 78202 69562 42205and 32396 respectively The worst case error of GA
1was
more than twice that of GA4Thus we would recommend the
combined algorithm GA4
6 Conclusions
In this paper we address a single-machine total completiontime problem with the sum of processing times basedlearning effect and release timesThe objective is to minimizethe total completion time The problem without learningconsideration is NP-hard one Thus we develop a branch-and-bound algorithm incorporatingwith several dominancesand two lower bounds to derive optimal solution and geneticalgorithms that was proposed to obtain near-optimal solu-tions respectively
The branch-and-bound algorithm performs well to solvethe instances of less than or equal to 24 jobs in a reasonableamount of time In the different varying parameter theproposed genetic algorithms are quite accurate for small sizeproblems and the mean error percentage of the proposedgenetic algorithms are less than 035 Another interestingtopic for future study is to consider and study the problem inthe multimachine environments or multicriteria cases
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 9
Table1Con
tinued
119899a
120579
Branch-a
nd-bou
ndalgorithm
GA
1GA
2GA
3GA
4Nod
eCP
Utim
eOF
Errorp
ercentage
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
Mean
Max
20
minus005
111989982
278
00156
00624
2016
268
74028
03538
69562
00113
02265
000
00000
00025
1319111
4407360
1059941
3224697
2007803
29510
05928
32815
08700
42205
02480
15709
050
963361
9983475
618895
5687017
2000377
040
1601608
1746
402224
15708
00033
00510
075
323776
2518001
229485
1831606
2000022
004
46000
00000
0000027
004
46000
00000
0010
041188
138955
33642
1110
7220
00017
00338
00017
00338
000
00000
00000
00000
00
minus010
11198991345
14123
01716
17317
2006707
34958
04690
28787
04353
34958
00000
00000
025
2134707
16978059
1543318
12044214
2003845
23175
01848
06798
03536
21979
004
6002909
050
202434
1320363
156266
1024927
2000000
00000
000
4500786
00000
00000
00000
00000
075
67733
450297
55645
3452
2920
000
00000
00000
00000
00000
00000
00000
00000
0010
030928
238389
26192
185483
2000023
004
62000
00000
0000023
004
62000
00000
00
minus015
1119899293
2224
004
2902964
2001670
13088
01845
29345
01835
29345
000
00000
00025
1539138
7544
291
1039052
4970
186
20046
8529738
01876
10582
02690
19134
00544
09353
050
235214
2737579
16906
61920840
20000
4600923
00032
006
4100073
01284
000
00000
00075
56901
281822
45716
225576
20000
00000
00000
00000
00000
00000
00000
00000
0010
017992
51529
15163
36973
20000
00000
00000
00000
00000
00000
00000
00000
00
minus020
1119899698
4825
00991
06240
2002436
15781
04239
404
2602224
15781
01927
15781
025
434800
664
668017
2750735
40004590
2001596
14118
00549
02968
02765
21869
00203
02249
050
107201
675811
80878
476895
20000
00000
00000
00000
0000058
01155
000
00000
00075
31529
163854
25990
110918
20000
00000
00000
00000
00000
00000
00000
00000
0010
015118
63845
13198
49453
20000
00000
0000016
00324
00016
00318
000
00000
00
24
minus005
1119899478
2082
01248
04836
2004707
39003
02017
20990
01368
18118
00959
18118
025
43205517
875044
4452338760
1060
43535
1304991
21418
06032
19032
046
0720265
02283
13231
050
11780220
66191783
13054869
65948027
1900758
10456
00313
02363
01334
10456
00161
02363
075
1216867
4261994
1562529
4708125
2000056
01126
00198
03359
00174
01871
000
00000
0010
0888475
5439067
1144057
69244
1720
000
00000
00000
00000
00000
00000
00000
00000
00
minus010
11198991074
12050
02496
26364
2007378
42545
03868
24023
02577
18340
00541
06057
025
24953166
5300
6630
28547095
6344
0156
1104324
22382
05843
29043
06883
25478
03452
22382
050
3116322
31612798
3661781
38159717
2000137
0118
400087
00883
00075
01028
000
00000
00075
513685
2435858
672926
2964331
20000
00000
0000015
00306
00112
01371
000
00000
0010
0270431
1462391
35260
91960620
20000
00000
0000052
01037
00039
00771
000
00000
00
minus015
1119899674
2613
01693
06084
2007280
78202
03115
27017
02512
34255
00284
05683
025
23417410
61951876
253876
736371360
415
09250
33429
02948
12902
02232
17855
00499
04354
050
4118014
36622393
4749373
39276216
2000114
0119
300053
01058
000
00000
00000
00000
00075
661765
5958864
874869
7920327
2000073
01458
00075
01509
000
4300867
000
00000
0010
0837932
11995859
968626
13244172
2000000
00000
00000
00000
00000
00000
00000
00000
minus020
11198993126
15554
07082
31980
2001942
32396
04290
33146
03196
32396
01669
32396
025
9333827
72575949
9993
219
80682031
1500589
02822
006
4607914
00844
07331
00204
02057
050
384242
2271989
508384
2808486
2000038
00760
00130
02596
00130
02596
00000
00000
075
570054
4053320
7017
325311990
20000
00000
00000
00000
00000
00000
00000
00000
0010
0233727
1202819
2915
7413260
0620
000
00000
00000
00000
00000
00000
00000
00000
00NoteldquoO
Frdquodeno
testhe
numbero
finstances
in20
setsof
datathatcanbe
solved
inlessthan
108no
desb
yusingtheb
ranch-
and-bo
undmetho
d
10 Mathematical Problems in Engineering
000
005
010
015
020
025
030
GA1 GA2 GA3 GA4GA
n16
n20n12
n24
Mea
n er
ror (
)
Figure 8 The performance of the genetic algorithms for various 119899
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
Mea
n er
ror (
)
minus005
minus010 minus020
minus015
Figure 9 The performance of the genetic algorithms for various 119886
percentage among all the 80 scenarios with varying learningrates numbers of jobs and values of 120579 The maximummean error percentages of GA
1 GA2 GA3 and GA
4were
16268 06032 08700 and 03452 respectively Themaximum mean error percentage of GA
1was more than
four times that of GA4 The combined algorithm GA
4also
clearly outperformed each of the three algorithms in terms
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
02505
0751
1n
Mea
n er
ror (
)
Figure 10The performance of the genetic algorithms for various 120579
of the maximum error percentage The worst cases of thecorresponding algorithms were 78202 69562 42205and 32396 respectively The worst case error of GA
1was
more than twice that of GA4Thus we would recommend the
combined algorithm GA4
6 Conclusions
In this paper we address a single-machine total completiontime problem with the sum of processing times basedlearning effect and release timesThe objective is to minimizethe total completion time The problem without learningconsideration is NP-hard one Thus we develop a branch-and-bound algorithm incorporatingwith several dominancesand two lower bounds to derive optimal solution and geneticalgorithms that was proposed to obtain near-optimal solu-tions respectively
The branch-and-bound algorithm performs well to solvethe instances of less than or equal to 24 jobs in a reasonableamount of time In the different varying parameter theproposed genetic algorithms are quite accurate for small sizeproblems and the mean error percentage of the proposedgenetic algorithms are less than 035 Another interestingtopic for future study is to consider and study the problem inthe multimachine environments or multicriteria cases
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
10 Mathematical Problems in Engineering
000
005
010
015
020
025
030
GA1 GA2 GA3 GA4GA
n16
n20n12
n24
Mea
n er
ror (
)
Figure 8 The performance of the genetic algorithms for various 119899
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
Mea
n er
ror (
)
minus005
minus010 minus020
minus015
Figure 9 The performance of the genetic algorithms for various 119886
percentage among all the 80 scenarios with varying learningrates numbers of jobs and values of 120579 The maximummean error percentages of GA
1 GA2 GA3 and GA
4were
16268 06032 08700 and 03452 respectively Themaximum mean error percentage of GA
1was more than
four times that of GA4 The combined algorithm GA
4also
clearly outperformed each of the three algorithms in terms
000
002
004
006
008
010
012
014
016
GA1 GA2 GA3 GA4GA
02505
0751
1n
Mea
n er
ror (
)
Figure 10The performance of the genetic algorithms for various 120579
of the maximum error percentage The worst cases of thecorresponding algorithms were 78202 69562 42205and 32396 respectively The worst case error of GA
1was
more than twice that of GA4Thus we would recommend the
combined algorithm GA4
6 Conclusions
In this paper we address a single-machine total completiontime problem with the sum of processing times basedlearning effect and release timesThe objective is to minimizethe total completion time The problem without learningconsideration is NP-hard one Thus we develop a branch-and-bound algorithm incorporatingwith several dominancesand two lower bounds to derive optimal solution and geneticalgorithms that was proposed to obtain near-optimal solu-tions respectively
The branch-and-bound algorithm performs well to solvethe instances of less than or equal to 24 jobs in a reasonableamount of time In the different varying parameter theproposed genetic algorithms are quite accurate for small sizeproblems and the mean error percentage of the proposedgenetic algorithms are less than 035 Another interestingtopic for future study is to consider and study the problem inthe multimachine environments or multicriteria cases
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 11
References
[1] D Biskup ldquoSingle-machine scheduling with learning consider-ationsrdquo European Journal of Operational Research vol 115 no 1pp 173ndash178 1999
[2] R G Vickson ldquoTwo single-machine sequencing problemsinvolving controllable job processing timesrdquo American Instituteof Industrial Engineers Transactions vol 12 no 3 pp 258ndash2621980
[3] E Nowicki and S Zdrzałka ldquoA survey of results for sequencingproblems with controllable processing timesrdquo Discrete AppliedMathematics vol 26 no 2-3 pp 271ndash287 1990
[4] T C E Cheng Z L Chen and L I Chung-Lun ldquoParallel-machine scheduling with controllable processing timesrdquo IIETransactions vol 28 no 2 pp 177ndash180 1996
[5] L Monch H Balasubramanian J W Fowler and M E PfundldquoHeuristic scheduling of jobs on parallel batch machines withincompatible job families and unequal ready timesrdquo Computersand Operations Research vol 32 no 11 pp 2731ndash2750 2005
[6] W-C Lee C-C Wu and M-F Liu ldquoA single-machine bi-criterion learning scheduling problem with release timesrdquoExpert Systems with Applications vol 36 no 7 pp 10295ndash103032009
[7] W-C Lee C-CWu and P-HHsu ldquoA single-machine learningeffect scheduling problem with release timesrdquo Omega vol 38no 1-2 pp 3ndash11 2010
[8] T Eren ldquoMinimizing the total weighted completion time ona single machine scheduling with release dates and a learningeffectrdquo Applied Mathematics and Computation vol 208 no 2pp 355ndash358 2009
[9] C-C Wu and C-L Liu ldquoMinimizing the makespan on a singlemachine with learning and unequal release timesrdquo Computersand Industrial Engineering vol 59 no 3 pp 419ndash424 2010
[10] M D Toksarı ldquoA branch and bound algorithm for minimizingmakespan on a singlemachinewith unequal release times underlearning effect and deteriorating jobsrdquo Computers amp OperationsResearch vol 38 no 9 pp 1361ndash1365 2011
[11] C-C Wu P-H Hsu J-C Chen and N-S Wang ldquoGeneticalgorithm for minimizing the total weighted completion timescheduling problem with learning and release timesrdquo Comput-ers amp Operations Research vol 38 no 7 pp 1025ndash1034 2011
[12] F Ahmadizar and L Hosseini ldquoBi-criteria single machinescheduling with a time-dependent learning effect and releasetimesrdquo Applied Mathematical Modelling vol 36 no 12 pp6203ndash6214 2012
[13] R Rudek ldquoComputational complexity of the single processormakespan minimization problem with release dates and job-dependent learningrdquo Journal of the Operational Research Soci-ety 2013
[14] D-C Li and P-H Hsu ldquoCompetitive two-agent schedulingwith learning effect and release times on a single machinerdquoMathematical Problems in Engineering vol 2013 Article ID754826 9 pages 2013
[15] J-Y Kung Y-P Chao K-I Lee C-C Kang and W-CLin ldquoTwo-agent single-machine scheduling of jobs with time-dependent processing times and ready timesrdquo MathematicalProblems in Engineering vol 2013 Article ID 806325 13 pages2013
[16] Y Yin W-H Wu W-H Wu and C-C Wu ldquoA branch-and-bound algorithm for a single machine sequencing tominimize the total tardiness with arbitrary release dates and
position-dependent learning effectsrdquo Information Sciences AnInternational Journal vol 256 pp 91ndash108 2014
[17] A H G Rinnooy Kan Machine Scheduling Problems Classifi-cations Complexity and Computations Martinus Nijhoff TheHague The Netherlands 1976
[18] J Heizer and B RenderOperations Management Prentice HallEnglewood Cliffs NJ USA 5th edition 1999
[19] T C E Cheng and G Wang ldquoSingle machine scheduling withlearning effect considerationsrdquo Annals of Operations Researchvol 98 no 1ndash4 pp 273ndash290 2000
[20] D Biskup ldquoA state-of-the-art review on scheduling with learn-ing effectsrdquo European Journal of Operational Research vol 188no 2 pp 315ndash329 2008
[21] J-B Wang C T Ng T C E Cheng and L L Liu ldquoSingle-machine scheduling with a time-dependent learning effectrdquoInternational Journal of Production Economics vol 111 no 2 pp802ndash811 2008
[22] W-H Kuo and D-L Yang ldquoMinimizing the total completiontime in a single-machine scheduling problem with a time-dependent learning effectrdquo European Journal of OperationalResearch vol 174 no 2 pp 1184ndash1190 2006
[23] T C E Cheng C-C Wu and W-C Lee ldquoSome schedul-ing problems with sum-of-processing-times-based and job-position-based learning effectsrdquo Information Sciences vol 178no 11 pp 2476ndash2487 2008
[24] C Koulamas and G J Kyparisis ldquoSingle-machine and two-machine flowshop scheduling with general learning functionsrdquoEuropean Journal of Operational Research vol 178 no 2 pp402ndash407 2007
[25] C-C Wu and W-C Lee ldquoSingle-machine and flowshopschedulingwith a general learning effectmodelrdquoComputers andIndustrial Engineering vol 56 no 4 pp 1553ndash1558 2009
[26] W-C Lee and C-C Wu ldquoSome single-machine and 119898-machine flowshop scheduling problems with learning consid-erationsrdquo Information Sciences vol 179 no 22 pp 3885ndash38922009
[27] Y Yin D Xu K Sun and H Li ldquoSome scheduling problemswith general position-dependent and time-dependent learningeffectsrdquo Information Sciences vol 179 no 14 pp 2416ndash24252009
[28] A Janiak and R Rudek ldquoExperience-based approach toscheduling problems with the learning effectrdquo IEEE Transac-tions on SystemsMan and Cybernetics A vol 39 no 2 pp 344ndash357 2009
[29] S-J Yang and D-L Yang ldquoA note on single-machine groupscheduling problems with position-based learning effectrdquoApplied Mathematical Modelling vol 34 no 12 pp 4306ndash43082010
[30] Y Yin D Xu and J Wang ldquoSingle-machine schedulingwith a general sum-of-actual-processing-times-based and job-position-based learning effectrdquo Applied Mathematical Mod-elling vol 34 no 11 pp 3623ndash3630 2010
[31] C-C Wu Y Yin and S-R Cheng ldquoSome single-machinescheduling problems with a truncation learning effectrdquo Com-puters and Industrial Engineering vol 60 no 4 pp 790ndash7952011
[32] J-B Wang andM-Z Wang ldquoA revision of Machine schedulingproblems with a general learning effectrdquo Mathematical andComputer Modelling vol 53 no 1-2 pp 330ndash336 2011
[33] Y-Y Lu C-M Wei and J-B Wang ldquoSeveral single-machinescheduling problems with general learning effectsrdquo AppliedMathematical Modelling vol 36 no 11 pp 5650ndash5656 2012
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
12 Mathematical Problems in Engineering
[34] J-B Wang and J-J Wang ldquoScheduling jobs with a generallearning effect modelrdquo Applied Mathematical Modelling vol 37no 4 pp 2364ndash2373 2013
[35] L Li S W Yang Y B Wu Y Huo and P Ji ldquoSingle machinescheduling jobs with a truncated sum-of-processing-times-based learning effectrdquo The International Journal of AdvancedManufacturing Technology vol 67 no 1ndash4 pp 261ndash267 2013
[36] T C E Cheng C-C Wu J-C Chen W-H Wu and S-RCheng ldquoTwo-machine flowshop scheduling with a truncatedlearning function to minimize the makespanrdquo InternationalJournal of Production Economics vol 141 no 1 pp 79ndash86 2013
[37] W-H Wu ldquoA two-agent single-machine scheduling problemwith learning and deteriorating considerationsrdquo MathematicalProblems in Engineering vol 2013 Article ID 648082 18 pages2013
[38] R L Graham E L Lawler J K Lenstra and A H GRinnooy Kan ldquoOptimization and approximation in determin-istic sequencing and scheduling a surveyrdquo Annals of DiscreteMathematics vol 5 pp 287ndash326 1979
[39] G H Hardy J E Littlewood and G Polya InequalitiesCambridge University Press London UK 1967
[40] J H Holland Adaptation in Natural and Artificial SystemsUniversity of Michigan Press Ann Arbor Mich USA 1975
[41] D E GoldbergGenetic Algorithms in Search Optimization andMachine Learning Addison-Wesley 1989
[42] M Gen and R Cheng Genetic Algorithms and EngineeringDesign John Wiley amp Sons New York NY USA 1996
[43] M Soolaki I Mahdavi N Mahdavi-Amiri R Hassanzadehand A Aghajani ldquoA new linear programming approach andgenetic algorithm for solving airline boarding problemrdquoAppliedMathematical Modelling vol 36 no 9 pp 4060ndash4072 2012
[44] S Fujita ldquoRetrieval parameter optimization using genetic algo-rithmsrdquo Information Processing and Management vol 45 no 6pp 664ndash682 2009
[45] W Fan M D Gordon and P Pathak ldquoA generic rankingfunction discovery framework by genetic programming forinformation retrievalrdquo Information Processing andManagementvol 40 no 4 pp 587ndash602 2004
[46] O Etiler B Toklu M Atak and J Wilson ldquoA genetic algorithmfor flow shop scheduling problemsrdquo Journal of the OperationalResearch Society vol 55 no 8 pp 830ndash835 2004
[47] M Nawaz E E Enscore Jr and I Ham ldquoA heuristic algorithmfor the m-machine n-job flow-shop sequencing problemrdquoOMEGA vol 11 no 1 pp 91ndash95 1983
[48] E Falkenauer and S Bouffouix ldquoA genetic algorithm for jobshoprdquo in Proceedings of the IEEE International Conference onRobotics and Automation pp 824ndash829 April 1991
[49] C Reeves ldquoHeuristics for scheduling a single machine subjectto unequal job release timesrdquo European Journal of OperationalResearch vol 80 no 2 pp 397ndash403 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of