mathematical comparison of defuzzification of fuzzy logic
TRANSCRIPT
© 2021, IJSRMSS All Rights Reserved 29
International Journal of Scientific Research in ___________________________ Research Paper. Mathematical and Statistical Sciences
Volume-8, Issue-2, pp.29-37, April (2021) E-ISSN: 2348-4519
Mathematical Comparison of Defuzzification of Fuzzy Logic Controller
for Intelligence Air Conditioning System
M.A. Islam1*
, M.S. Hossain2, I.S.M. Haque
3
1,2,3
Department of Mathematics, University of Rajshahi, Rajshahi-6205, Bangladesh
*Corresponding Author: [email protected], Tel.: +8801773387844
Available online at: www.isroset.org
Received: 27/Feb/2021, Accepted: 07/Apr/2021, Online: 30/Apr/2021
Abstract—The objective of this paper is to compare several defuzzification techniques for the application of an air cooler
control design in order to predict the error difference between different methods. The four input parameters are considered
to be user temperature (UT), temperature difference (Tdf), dew point (Td), and number of peoples (NP). However,
Compressor Speed (CS), Fan Speed (FS) and Fin Direction (FD) are used as outputs. The simulation of this paper is carried
out by using MATLAB.
Keywords—Fuzzy logic controller, defuzzification techniques, fuzzy rule base, MATLAB simulation.
1. INTRODUCTION
In 1965, the idea of Fuzzy Logic was proposed by L.A.
Zadeh in one of his research papers under the name "Fuzzy
Logic or Fuzzy Sets"[1, 2]. In 1974, Mamdani [3] applied
the fuzzy logic in a practical application to control an
automatic steam engine. Today, Fuzzy Logic makes a
tremendous change in the development of uncertainty in
artificial intelligence system. However, it is easy to
implement and its impact in modern science is very ground
breaking. Its main objective is to simulate human brain on
decision making mechanism in intelligence system while
working in a vague, precise or uncertain environment.
Although it doesn’t need total knowledge about the
characteristic of the model, it helps machine to think like
human and respond instantly. The contribution of FIS
(Fuzzy Inference System) in this manner is not doubtful.
Its working principle can be divided into four mechanisms
like Fuzzification, Fuzzy Rules Based System, Fuzzy
Inference Engine and Defuzzification [4]. With the
introduction of linguistic variable, it is easy to describe
fuzzy knowledge-based system where its transition is very
smooth between rules-based systems. Besides,
Fuzzification and Defuzzification system refer to the
process of making fuzzy and crisp output. In order to get
the target output of an air conditioner by manipulating
temperature and humidity to save the energy of compressor
and fan while using all the resources, fuzzy logic rules-
based system is used for fuzzification, defuzzification for
the acquiring the target output [5]. In 2012, S. K, Dash et
al. [6] proposed an automated intelligent air conditioning
system while utilizing all available resources including
climatic condition in the efficient manner in order to
provide the user comfortable cooling level and also
optimized energy consumption. In 2015, S. M. Sabhy et al.
[7] developed fuzzy logic Control system of an air
conditioner where he used several linguistic variables and
showed how membership function works in air
conditioning system. Another researcher A. Saepullah [1]
modified FIS to save the energy of air conditioner by using
Mamdani, Sugeno, and Tsukamoto method in 2015.
Further, R. Roshmi et al. [8], S. M. Sofi et al. [9] Rajkumar
et al. [10], R. Kiruthika et al. [11], A. Sandanasamy et al.
[12] and Mohammad Sammany et al. [13] are also working
on fuzzy.
Our aim is to develop fuzzy logic Control System, in order
to make it automated by using triangular and trapezoidal
fuzzy number. This paper provides the idea of improving
air conditioner’s air quality. Here, we demonstrated a little
bit about FIS and in the introductory section while the
simulation results and graphical representation which was
obtained by using MATLAB are discussed afterwards.
Also, this paper describes how automated air conditioning
system can be improved which is based on the imprecise
input sensors.
2. AIR ORGANIZATION SYSTEM
Due point temperature indicates both Temperature and
Relative Humidity (RH) in a certain place. In a certain due
point temperature, if RH increases than Temperature will
decrease. Again, if Temperature increases than RH will
decrease. By controlling due point, material’s decay can be
decreased. Moreover, it is used to measure humidity. The
table 1 shown below shows human’s reaction on a standard
dew point.
Table 1. A standard Dew Point Human Reaction Table
Dew point (Td) Human Reaction
Less than 8° Very Dry
8°- 14° Dry
14°-18° Refreshing
19°-20° Humid
20°-25° Comfortable
Above 25° Painful
Int. J. Sci. Res. in Mathematical and Statistical Sciences Vol. 8, Issue.2, 2021
© 2021, IJSRMSS All Rights Reserved 30
Air Condition Fuzzy Logic control system takes four
variables into consideration showing in the following
block:
(1) User temperature (14°C-30° continuous control).
(2) Temperature difference (Tdf).
(3) Dew point (Td).
(4) Number of people (NP).
User temperature subtracted from temperature difference
before sending data for fuzzification step. Fuzzy arithmetic
and criterion step is applied on these variables and final
result is defuzzified step to get following crisp results as
showing in the following figure 1.
(1) Compressor Speed (CS).
(2) Fan Speed (FS).
(3) Fin Direction (FD).
Figure 1. FLC for Air Conditioning System.
3. FUZZY LOGIC CONTROLLER (FLC)
If-Then rules-based format is solved by fuzzy logic
controller in four steps as follows:
Figure 2. Fuzzy Logic Controller (FLC) System
Step-1 Input Variables: Firstly, a set of MF of linguistic
variables are taken as inputs where input variables are
basically words or sentences. In order to take any fuzzy or
crisp output by fuzzification or defuzzification respectably,
fuzzy MF are used.
Step-2 Fuzzification: Fuzzification is a process which
provides fuzzy output. Here, crisp values are fuzzified for
fuzzy output.
Step-3 Fuzzy Inference System: Fuzzy rule-based is
applied here. For each antecedent, there is a consequent
which is the resulted output for every rule-base. Several
operators like or, and, else, not are used in rule-based
system to connect multiple linguistic variables so that
target output is obtained inside the inference system.
Step-4 Defuzzification: In the defuzzification process,
crisp output is obtained from fuzzy set. Hardware
applications are greatly dependent on defuzzification
system.
4. DEFUZZIFICATION TECHNIQUES
In general, the fuzzification process involves the union of
two or more fuzzy sets, noted as, input sets” and the sets
are defined as the universe of discourse. This process
generates outputs from the precise values of input which
are in fuzzy form. The outputs of the fuzzification process
depend on the rule base and inference engine.
This process of converting the fuzzy rule based output to
crisp value for the designed system is known as
defuzzification. The defuzzification method cannot be
chosen systematically following the applications. It
depends on the need of the application.
4.1 CENTROID METHOD The "centroid method" also known as "center of gravity" or
"center of area" method is a technique which is most
commonly employed and familiar for defuzzification. It
reduces the area to smaller regions and a combined
operation is performed to obtain the final output. It is
stated by the expression below [14]:
1
1
.n
i ii
n
ii
xx x
x
Here, n represents the number of elements in the sample,
i.e., ix are the elements and ix are their
corresponding membership functions.
4.2 BISECTOR METHOD The area is divided into two regions by the bisector method
which may or may not coincide with the centroid line of
the given area.
4.3 MEAN OF MAXIMA (MOM) METHOD Actually, the easiest way of defuzzification is to take the
nearby crisp value with highest membership function. The
arithmetic average of all mean values of the intervals that
contains fuzzy set including zero length intervals is called
mean of maxima method. General equation of this method
is given by [14]
Int. J. Sci. Res. in Mathematical and Statistical Sciences Vol. 8, Issue.2, 2021
© 2021, IJSRMSS All Rights Reserved 31
| |
i
iMxxM
x
Where M height of the fuzzy set and
| |M Cardinality of the fuzzy set .M
5. FUZZY MEMBERSHIP FUNCTION
Membership function is used to represent a fuzzy set
graphically. Triangular and trapezoidal MF (Membership
Function) are used in FIS. In this paper, we will use
triangular and trapezoidal membership function to develop
Fuzzy Logic Controller (FLC). Each input and output
including their linguistic variable are discussed below.
5.1 FUZZY INPUT VARIABLES
5.1.1 User Temperature (UT)
From the several sensors like thermostat or electronic,
user’s temperature can be collected which has three
linguistic variables named low, medium and high as shown
in table 2.
Table 2: User Temperature (UT) Classification
Range Fuzzy Set
User Temperature (UT) 14-26 Low
22-28 Medium
26-34 High
5.1.2 Temperature Difference (Tdf)
The difference between user’s temperature and room’s
temperature is the temperature difference (Tdf). In this
paper, the range of Tdf is taken between -4 to +4. As soon
as the difference goes out of the range, the air conditioner
switched off since it can’t be worked as a heat pump. Four
linguistic variables are used here which are neg (negative),
zero, pos (positive), hpos (high positive) as shown in table
3.
Table 3: Temperature Difference (Tdf) Classification
Range Fuzzy Set
-4 – 0 neg
Temperature Difference (Tdf) -1.5 - +1.5 zero
0 – +2.5 pos
+2 - +4 hpos
5.1.3 Dew Point (Td)
Dew point temperature is the temperature of captive room
where air conditioner is placed. Here, two linguistic
variables are taken named Low and High as shown in the
table 4.
Table 4: Dew Point (Td) Classification
Range Fuzzy Set
Dew Point (Td) 12-20 Low
17-29 High
5.1.4 Number of People (NP)
The number of people who are staying inside the room are
the number of people. If there is no one inside the room
then the air conditioner will remain off. Three linguistic
variables are taken here which are Low, Medium and High
as shown in the table 5.
Table 5: Number of People (NP) Classification
Range Fuzzy Set
Number of People (NP) 1-4 Low
3-9 Medium
7-12 High
5.2 FUZZY OUTPUT VARIABLES
5.2.1 Compressor Speed (CS)
Actually, the compressor speed varies from 0 to 100%
depending on the inputs that are taken by the sensors from
the room. By controlling the compressor speed, room
temperature can be optimized. Three linguistic variables
are taken here which are Slow, Medium and Fast as shown
in the table 6.
Table 6: Compressor Speed (CS) Classification
Range Fuzzy Set
Compressor Speed (CS) 0-55 Slow
45-75 Medium
70-100 Fast
5.2.2 Fan Speed (FS)
Inside an air conditioner fan is running with a speed that
varies from 0 to 100% depending on the inputs. Here,
Slow, Medium and Fast are taken as linguistic variables as
shown in the table 7.
Table 7: Fan Speed (FS) Classification
Range Fuzzy Set
Fan Speed (FS) 0-50 Slow
45-80 Medium
70-100 Fast
5.1.1 Fin Direction (FD))
The fin direction indicates the direction of the cold air flow
that comes from the fin which is a bunch of blades
bounded to the air conditioner. Its direction is either
towards (Td) the user or away (Aw) from the user.
Table 8: Fin Direction (FD) Classification
Range Fuzzy Set
Fin Direction (FD) 0-50 Slow
45-80 Medium
70-100 Fast
Here in the Fig. 3 to 9 inputs and outputs membership
functions of fuzzy logic controller shown:
Fig. 3: Input of FLC “User Temperature (UT)”
Int. J. Sci. Res. in Mathematical and Statistical Sciences Vol. 8, Issue.2, 2021
© 2021, IJSRMSS All Rights Reserved 32
Fig. 4: Input of FLC “Temperature Difference (Tdf)”
Fig. 5: Input of FLC “Dew Point (Td)”
Fig. 6: Input of FLC “Number of People (NP)”
Fig. 7: Output of FLC “Compressor Speed (CS)”
Fig. 8: Output of FLC “Fan Speed (FS)”
Fig. 9: Output of FLC “Fin Direction (FD)”
6. FUZZY RULE BASE
Fuzzy Rules based are applied in the FLC by selecting the
appropriate sequence in the IF-Then rules which are based
on natural language. It is designed to make any automated
decision. It is formed by keeping the relationship among
Input and output in mind. The input variables (UT, Tdf,
Td, and NP) make a total of 3*4*2*3=72 rules as shown in
the followings:
1. If (UT is Low) and (Tdf is neg) and (Td is Low)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
2. If (UT is Low) and (Tdf is zero) and (Td is Low)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
3. If (UT is Low) and (Tdf is pos) and (Td is Low)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
4. If (UT is Low) and (Tdf is hpos) and (Td is Low)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
5. If (UT is Low) and (Tdf is neg) and (Td is Low)
and (NP is Medium) then (CS is Slow) (FS is
Slow) (FD is towards).
6. If (UT is Low) and (Tdf is zero) and (Td is Low)
and (NP is Medium) then (CS is Slow) (FS is
Slow) (FD is towards).
7. If (UT is Low) and (Tdf is pos) and (Td is Low)
and (NP is Medium) then (CS is Slow) (FS is
Medium) (FD is away).
Int. J. Sci. Res. in Mathematical and Statistical Sciences Vol. 8, Issue.2, 2021
© 2021, IJSRMSS All Rights Reserved 33
8. If (UT is Low) and (Tdf is hpos) and (Td is Low)
and (NP is Medium) then (CS is Medium) (FS is
Medium) (FD is away).
9. If (UT is Low) and (Tdf is neg) and (Td is Low)
and (NP is High) then (CS is Slow) (FS is Slow)
(FD is away).
10. If (UT is Low) and (Tdf is zero) and (Td is Low)
and (NP is High) then (CS is Slow) (FS is Slow)
(FD is towards).
11. If (UT is Low) and (Tdf is pos) and (Td is Low)
and (NP is High) then (CS is Medium) (FS is
Medium) (FD is away).
12. If (UT is Low) and (Tdf is hpos) and (Td is Low)
and (NP is High) then (CS is Fast) (FS is Fast)
(FD is away).
13. If (UT is Low) and (Tdf is neg) and (Td is High)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
14. If (UT is Low) and (Tdf is zero) and (Td is High)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
15. If (UT is Low) and (Tdf is pos) and (Td is High)
and (NP is Low) then (CS is Medium) (FS is
Slow) (FD is away).
16. If (UT is Low) and (Tdf is hpos) and (Td is High)
and (NP is Low) then (CS is Slow) (FS is
Medium) (FD is away).
17. If (UT is Low) and (Tdf is neg) and (Td is High)
and (NP is Medium) then (CS is Slow) (FS is
Slow) (FD is away).
18. If (UT is Low) and (Tdf is zero) and (Td is High)
and (NP is Medium) then (CS is Slow) (FS is
Slow) (FD is towards).
19. If (UT is Low) and (Tdf is pos) and (Td is High)
and (NP is Medium) then (CS is Slow) (FS is
Medium) (FD is away).
20. If (UT is Low) and (Tdf is hpos) and (Td is High)
and (NP is Medium) then (CS is Medium) (FS is
Medium) (FD is away).
21. If (UT is Low) and (Tdf is neg) and (Td is High)
and (NP is High) then (CS is Slow) (FS is Slow)
(FD is away).
22. If (UT is Low) and (Tdf is zero) and (Td is High)
and (NP is High) then (CS is Slow) (FS is Slow)
(FD is towards).
23. If (UT is Low) and (Tdf is pos) and (Td is High)
and (NP is High) then (CS is Fast) (FS is Fast)
(FD is away).
24. If (UT is Low) and (Tdf is hpos) and (Td is High)
and (NP is High) then (CS is Fast) (FS is Fast)
(FD is away).
25. If (UT is Medium) and (Tdf is neg) and (Td is
Low) and (NP is Low) then (CS is Slow) (FS is
Slow) (FD is towards).
26. If (UT is Medium) and (Tdf is zero) and (Td is
Low) and (NP is Low) then (CS is Slow) (FS is
Slow) (FD is towards).
27. If (UT is Medium) and (Tdf is pos) and (Td is
Low) and (NP is Low) then (CS is Slow) (FS is
Slow) (FD is towards).
28. If (UT is Medium) and (Tdf is hpos) and (Td is
Low) and (NP is Low) then (CS is Slow) (FS is
Slow) (FD is towards).
29. If (UT is Medium) and (Tdf is neg) and (Td is
Low) and (NP is Medium) then (CS is Slow) (FS
is Slow) (FD is towards).
30. If (UT is Medium) and (Tdf is zero) and (Td is
Low) and (NP is Medium) then (CS is Slow) (FS
is Slow) (FD is towards).
31. If (UT is Medium) and (Tdf is pos) and (Td is
Low) and (NP is Medium) then (CS is Slow) (FS
is Medium) (FD is away).
32. If (UT is Medium) and (Tdf is hpos) and (Td is
Low) and (NP is Medium) then (CS is Medium)
(FS is Fast) (FD is away).
33. If (UT is Medium) and (Tdf is neg) and (Td is
Low) and (NP is High) then (CS is Slow) (FS is
Slow) (FD is towards).
34. If (UT is Medium) and (Tdf is zero) and (Td is
Low) and (NP is High) then (CS is Slow) (FS is
Slow) (FD is towards).
35. If (UT is Medium) and (Tdf is pos) and (Td is
Low) and (NP is High) then (CS is Medium) (FS
is Medium) (FD is away).
36. If (UT is Medium) and (Tdf is hpos) and (Td is
Low) and (NP is High) then (CS is Fast) (FS is
Fast) (FD is away).
37. If (UT is Medium) and (Tdf is neg) and (Td is
High) and (NP is Low) then (CS is Slow) (FS is
Slow) (FD is towards).
38. If (UT is Medium) and (Tdf is zero) and (Td is
High) and (NP is Low) then (CS is Slow) (FS is
Slow) (FD is towards).
39. If (UT is Medium) and (Tdf is pos) and (Td is
High) and (NP is Low) then (CS is Medium) (FS
is Slow) (FD is away).
40. If (UT is Medium) and (Tdf is hpos) and (Td is
High) and (NP is Low) then (CS is Medium) (FS
is Medium) (FD is away).
41. If (UT is Medium) and (Tdf is neg) and (Td is
High) and (NP is Medium) then (CS is Slow) (FS
is Slow) (FD is towards).
42. If (UT is Medium) and (Tdf is zero) and (Td is
High) and (NP is Medium) then (CS is Slow) (FS
is Slow) (FD is towards).
43. If (UT is Medium) and (Tdf is pos) and (Td is
High) and (NP is Medium) then (CS is Medium)
(FS is Medium) (FD is away).
44. If (UT is Medium) and (Tdf is hpos) and (Td is
High) and (NP is Medium) then (CS is Medium)
(FS is Fast) (FD is away).
45. If (UT is Medium) and (Tdf is neg) and (Td is
High) and (NP is High) then (CS is Slow) (FS is
Slow) (FD is towards).
46. If (UT is Medium) and (Tdf is zero) and (Td is
High) and (NP is High) then (CS is Medium) (FS
is Medium) (FD is away).
47. If (UT is Medium) and (Tdf is pos) and (Td is
High) and (NP is High) then (CS is Fast) (FS is
Fast) (FD is away).
Int. J. Sci. Res. in Mathematical and Statistical Sciences Vol. 8, Issue.2, 2021
© 2021, IJSRMSS All Rights Reserved 34
48. If (UT is Medium) and (Tdf is hpos) and (Td is
High) and (NP is High) then (CS is Fast) (FS is
Fast) (FD is away).
49. If (UT is High) and (Tdf is neg) and (Td is Low)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
50. If (UT is High) and (Tdf is zero) and (Td is Low)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
51. If (UT is High) and (Tdf is pos) and (Td is Low)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
52. If (UT is High) and (Tdf is hpos) and (Td is Low)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
53. If (UT is High) and (Tdf is neg) and (Td is Low)
and (NP is Medium) then (CS is Slow) (FS is
Slow) (FD is towards).
54. If (UT is High) and (Tdf is zero) and (Td is Low)
and (NP is Medium) then (CS is Slow) (FS is
Medium) (FD is away).
55. If (UT is High) and (Tdf is pos) and (Td is Low)
and (NP is Medium) then (CS is Medium) (FS is
Medium) (FD is away).
56. If (UT is High) and (Tdf is hpos) and (Td is Low)
and (NP is Medium) then (CS is Fast) (FS is
Medium) (FD is away).
57. If (UT is High) and (Tdf is neg) and (Td is Low)
and (NP is High) then (CS is Slow) (FS is Slow)
(FD is towards).
58. If (UT is High) and (Tdf is zero) and (Td is Low)
and (NP is High) then (CS is Slow) (FS is Slow)
(FD is towards).
59. If (UT is High) and (Tdf is pos) and (Td is Low)
and (NP is High) then (CS is Medium) (FS is
Medium) (FD is away).
60. If (UT is High) and (Tdf is hpos) and (Td is Low)
and (NP is High) then (CS is Fast) (FS is Fast)
(FD is away).
61. If (UT is High) and (Tdf is neg) and (Td is High)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
62. If (UT is High) and (Tdf is zero) and (Td is High)
and (NP is Low) then (CS is Slow) (FS is Slow)
(FD is towards).
63. If (UT is High) and (Tdf is pos) and (Td is High)
and (NP is Low) then (CS is Slow) (FS is
Medium) (FD is away).
64. If (UT is High) and (Tdf is hpos) and (Td is High)
and (NP is Low) then (CS is Medium) (FS is
Fast) (FD is away).
65. If (UT is High) and (Tdf is neg) and (Td is High)
and (NP is Medium) then (CS is Slow) (FS is
Slow) (FD is towards).
66. If (UT is High) and (Tdf is zero) and (Td is High)
and (NP is Medium) then (CS is Slow) (FS is
Slow) (FD is towards).
67. If (UT is High) and (Tdf is pos) and (Td is High)
and (NP is Medium) then (CS is Fast) (FS is
Medium) (FD is away).
68. If (UT is High) and (Tdf is hpos) and (Td is
High) and (NP is Medium) then (CS is Fast) (FS
is Fast) (FD is away).
69. If (UT is High) and (Tdf is neg) and (Td is High)
and (NP is High) then (CS is Slow) (FS is Slow)
(FD is towards).
70. If (UT is High) and (Tdf is zero) and (Td is High)
and (NP is High) then (CS is Medium) (FS is
Fast) (FD is away).
71. If (UT is High) and (Tdf is pos) and (Td is High)
and (NP is High) then (CS is Fast) (FS is Fast)
(FD is away).
72. If (UT is High) and (Tdf is hpos) and (Td is
High) and (NP is High) then (CS is Fast) (FS is
Fast) (FD is away).
7. FUZZY LOGIC CONTROLLER (FLC) IMP-
PLEMENTATION IN AIR CONDITIONING
USING MATLAB
7.1 FUZZY BASE CLASS
Mamdani method is used to create system control rules
obtained from experienced human operators [15]. In this
paper, Mamdani method is used to illustrate and centroid
method is used for defuzzification. Here, FIS editor FIS
Editor defines the Fuzzy Base Class, the various inputs,
i.e.User temperature (UT), Temperature Difference (Tdf),
Dew Point (Td), and Number of People (NP) and the
various output variables like Compressor Speed (CS), Fan
Speed (FS) and, Fin Direction (FD) [16] as shown.
Fig. 10: Fuzzy Base Class
7.2 FUZZY RULE BASE
By a user or deliberately, the rule of fuzzy can be planned
manually, for all situations the rule editor propagates rules
of the marked input variable and a user fills consequent
fuzzy items. The illustration of inputs and the fuzzy
outputs depicted on the basis of [16] as shown.
Int. J. Sci. Res. in Mathematical and Statistical Sciences Vol. 8, Issue.2, 2021
© 2021, IJSRMSS All Rights Reserved 35
Fig. 12: Fuzzy Base Rules
7.3 SOURFACE PLOTS
By applying three defuzzification methods like- centroid
method, bisector method and mean of maxima method. We
get the surface plots which are shown as [17] based on the
above effectuation.
Fig: 13 Surface plots for FLC of Compressor Speed using the
Centroid method
Fig. 14: Surface plot for FLC of Fan Speed using the Centroid
method
Fig. 15: Surface plot for FLC of Fin Direction using the
Centroid method
Fig. 16: Surface plot for FLC of Compressor Speed using
the Bisector method
Fig. 17: Surface plot for FLC of Fan Speed using the
Bisector method
Fig. 18: Surface plot for FLC of Fin Direction using the Bisector
method
Fig. 19: Surface plot for FLC of Compressor Speed using the
MOM method
Fig. 20: Surface plot for FLC of Fan Speed using the MOM
method
Fig. 21: Surface plot for FLC of Fin Direction using the MOM
method
Int. J. Sci. Res. in Mathematical and Statistical Sciences Vol. 8, Issue.2, 2021
© 2021, IJSRMSS All Rights Reserved 36
8.RESULT AND DISCUSSION
Based on the user's temperature, temperature difference,
dew point and number of people it can be said that this
method will have different compressor speed, fan speed
and fin direction. Here, a large number of values of user of
temperature, temperature difference, dew point and number
of people can be used to evaluate the difference between
these methods. Clearly it can be said that, these three
techniques shown in the above table which is used for
comparison, provides compressor speed, fan speed and fin
direction as output. Hence, by using “if” and “then” rules
during the defuzzification of the air conditioning system,
it’s output can be modified and also verified the same
result. Any modification in the rule base model will require
modify the if” and “then” rules to get a correct output.The
compressor speed, fan speed and fin direction outputs for
varying input parameters as shown in the tables 9, 10 and
11 respectively.
Table 9: Compressor Speed output for varying input parameters
UT
Tdf
Td
NP
Compressor Speed (CS)
Centroid Bisector MOM
24 0 20 6 28.1 28 28.5
22 2 24 7 39.1 33 29.5
24 1 25 5 39.3 40 60
Table 10: Fan Speed output for varying input parameters
UT
Tdf
Td
NP
Fan Speed (FS)
Centroid Bisector MOM
24 0 20 6 28.1 28 28.5
22 2 24 7 69.5 67 64.5
24 1 25 5 44.9 49 64.5
Table 11: Fin Direction output for varying input parameters
UT
Tdf
Td
NP
Fin Direction (FD)
Centroid Bisector MOM
24 0 20 6 36.2 36 37.4
22 2 24 7 60 60.3 59.5
24 1 25 5 49.1 52.2 59.5
For compressor speed, we can see if Dew point increases
then compressor speed, increases drastically.
Hence, mathematically center of gravity method (COG) is
the best option for this sequence. Now, for Fan Speed, if
Dew point and Number of people increases then Fan Speed
also increase at the same time where bisector method is the
best option by our observation.
Finally, for Fin Direction, we can see if Temperature
difference and Number of people increases then Fin
Direction also increases sequentially. Here, we can use one
of these two method (bisector, center of gravity) for the
best output.
Fig. 22: Defuzzification output for FLC using the Centroid
method
Fig. 23: Defuzzification output for FLC using the Bisector
method
Fig. 24: Defuzzification output for FLC using the MOM method
9.CONCLUSION
From the above discussion, we can clearly see that, it is
better to use Center of gravity method and bisector method
for the defuzzification of Compressor Speed and Fan
Speed respectively. However, for the defuzzification of Fin
Direction, we can use bisector method or Center of gravity
method on our own interest.
ACKNOWLEDGMENT
Authors of this paper are thankful to other authors whose
names are included in the references section for their
suggestions which helps us to modify this paper.
REFERENCES
[1] C. Elmas, Bulanik Mantik Denetleyicleri, ISBN 975-347-613,
2003.
[2] L. A. Zadeh, “Fuzzy Sets”, Information and Control, Vol.8,
pp.338-353, 1965.
[3] E. H. Mamdani, “Application of Fuzzy Algorithm for Control of
Simple Dynamic Plant”, Proc. IEEE, Vol.121, Issue.12,
pp.1585-1588, 1974. [4] J. Jantzen, “Tutorial on fuzzy logic”, Technical University of
Denmark, Dept. of Automation, Technical Report, 1998.
[5] N. Wakami, S. Araki, H. Nomura, “Recent applications of fuzzy
logic to home appliances”, in Proceedings of IECON’93-19th
Annual Conference of IEEE Industrial Electronics. IEEE
pp.155–160, 1993.
[6] S. K. Dash, G. Mohanty, A. Mohanty, “Intelligent air
conditioning system using fuzzy logic”, International Journal of
Scientific & Engineering Research, Vol.3, Issue.12, pp.1-6,
2012 [7] S. M. Sobhy, W. M. Khedr, “Developing of fuzzy logic
controller for air condition system”, International Journal of
Computer Applications, Vol.126, Issue.15, pp.1-8, 2015.
[8] R. Roshmi, M.S. Hossain, “Regularity and Normality in
Bitopological Spaces”, Int. J. Sci. Res. In Mathematical and
Statistical Sciences, Vol.7, Issue.1, pp.21-25, 2020.
Int. J. Sci. Res. in Mathematical and Statistical Sciences Vol. 8, Issue.2, 2021
© 2021, IJSRMSS All Rights Reserved 37
[9] S. M. Sofi, S. Peerzada, M. A. K. Baig, “Parametric
Generalizations of ‘Useful’R-Norm Fuzzy Information
Measures”, Int. J. Sci. Res. In Mathematical and Statistical
Sciences, Vol.5, Issue.6, pp.164-169, 2018.
[10] Rajkumar , S. Udayakumar, “Multicriterion Fuzzy Decision
Making in Irrigation Planning”, Int. J. Sci. Res. In
Mathematical and Statistical Sciences, Vol.6, Issue.2, pp.325-
333, 2019. [11] R. Kiruthika, V. Vijaykumar, “An Enhanced Fuzzy Based
Linkage Clustering Algorithm (EFCA) in High Dimensional
Data”, International Journal of Computer Sciences and
Engineering, Vol.8, Issue.2, pp.12-17, 2020.
[12] A. Sandanasamy, R. Thamarai Selvi, “An Quality Based
Software Requirement Prioritization Using Takagi-Sugeno
Fuzzy Inference”, International Journal of Computer Sciences
and Engineering, Vol.7, Issue.5, pp.642-647, 2019.
[13] Mohammad Sammany, Ahmad Steef, Nedaa Agami, T. Medhat,
“Fuzzy Mathematical Model For Optimizing Success Criteria
Of Projects: A Project Management Application,” International
Journal of Scientific Research in Computer Science and
Engineering, Vol.8, Issue.5, pp.77-83, 2020.
[14] H. Han, Chun-Yi, Yury. S, “Adaptive Control of a Class of
Non-linear Systems with Non-Linearly Parameterized Fuzzy
Approximation”, IEEE Transaction on Fuzzy Systems, Vol.9,
Issue.2, pp.315-323, 2001. [15] E. H. Mamdani, S. Assilian, “An experiment in linguistic
synthesis with a fuzzy logic controller”, International journal of
man-machine studies, Vol.7, Issue.1, pp.1-13, 1975.
[16] L. A. Zadeh , C. Berkeley, “Fuzzy logic toolbox”, Math Works,
2001. [17] S. Naaz, A. Alam, R. Biswas, “Effect of different
defuzzification methods in a fuzzy based load balancing
application”, IJCSI International Journal of Computer Science,
Vol.8, Issue.5, pp.261, 2011.
AUTHORS PROFILE
Md. Azharul Islam was born in
Rangpur, Bangladesh, in 1995. He
received his B.Sc. and M.Sc. degree in
Mathematics from University of
Rajshahi, Faculty of Science,
Bangladesh, in 2017 and 2018,
respectively. His research interest
includes Fuzzy Algebra, Fuzzy
Relation, Fuzzy Optimization, Fuzzy logic, Fuzzy
inference, system, Fuzzy control system, Machine learning
and networked control system, Artificial Intelligence (AI).
Md. Sahadat Hossain was born in
Rangpur, Bangladesh, in 1977. He
obtained his M.Sc. in 1999, M.Phil. in
2005, Ph. D. in 2012, in Fuzzy
Bitopology, from Rajshahi University,
Department of Mathematics, Faculty
of Science. Currently, he is employed
at the same university in the
department of Mathematics as a professor. He is the author
of 1 book and more than 40 other publications, including
more than 15 journal articles. His research interest includes
Fuzzy Algebra, Fuzzy Topology, Fuzzy logic, Fuzzy
control system, Fuzzy algorithm, and Fuzzy inference
system.
Ibna Sina Munzurul Haque was born
in 1995 in Meherpur, Khulna,
Bangladesh. He received his B.Sc.
degree in Mathematics from the
University of Rajshahi, Bangladesh in
2018. His research interests are Fuzzy
Logic, Fuzzy Algebra, Fuzzy Inference
System, Fuzzy Control System, Fuzzy
relation, Artificial Intelligence (Ai), Neural Network
(ANN) and, Machine learning (ML).