relative approach of firing angle optimization scheme in ... · generation of inexhaustible power...
TRANSCRIPT
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 13, Number 7 (2018) pp. 5057-5064
© Research India Publications. http://www.ripublication.com
5057
Relative Approach of Firing Angle Optimization Scheme in Grid Connected
7 Level CHB-MLI
Shazma Khan1, Balvinder Singh2 and Pratibha Singh3
1M.Tech Scholar, Power System 2,3Assistant Professor, Department of Electrical Engineering, Govt. Women Engineering College Ajmer, India.
Abstract
This paper presents the execution of two Artificial Intelligence
(AI) techniques- Artificial Neural Network (ANN) and Fuzzy
Logic (FL) in the PV fed Cascaded Multi-Level Inverter
(CMLI). Condition of voltage variance are arises in the PV
system owing to variable irradiance or temperature. In this
paper ANN and FL are implemented to determine the
optimized value of firing angles for unequal and equal dc
voltages of seven level CMLI such that harmonics will be
eliminated in the system. ANN and FL are technologies of
almost same age. In this paper we have tried to train our AI
controller from the data obtain by using Genetic Algorithm.
Due to partial shading voltage variation of about 10-20% is
observed and here approximately 20% voltage variance are
taken. THD attained by implementing ANN and FL are
displayed in the look-up table and MATLAB/SIMULINK
domain is used to validate the simulation result.
Keywords: Artificial Neural Network (ANN); Fuzzy Logic
(FL); Genetic Algorithm (GA); Multi-Level Inverter (MLI);
Total Harmonics Distortion (THD).
INTRODUCTION
In present scenario, there is an immediate requirement to
accelerate the evolution of leading energy skills in order to
address the universal deviations of sustainable expansion,
variation in climate and green energy. Generation of
inexhaustible power helps the nations to come across these
challenges and hence re-organize the prevailing system.
Photovoltaic energy is a commercially feasible and consistent
tool with huge potential in the upcoming years.
Recently, AI techniques are gaining importance as an
alternative to conventional practices [1]. These are engaged to
resolve complex issues of real-world in numerous regions for
example in handwriting recognition, engineering zones,
robotics, computer numerical control etc. Nowadays these
techniques gaining importance due to their attractive features.
These are - capacity to handle noisy data as well as incomplete
data; it can learn from samples; it can handle non-linearity;
handle complex issues; and when trained can do estimation and
simplification quickly.
AI implemented system are established and deployed globally
in various purpose as mentioned. In this paper, AI-techniques
are employed in order to size the PV linked Cascaded H-Bridge
Multi Level Inverter (CHB-MLI) systems. Foremost benefit of
applying an AI technique in the sizing of PV fed systems is its
noble optimization, wherever data are not all the time
accessible [2][3].
AI methods are classified as: Fuzzy Logic, Artificial Neural
Network, Genetic Algorithm, wavelets, hybrid for example
ANN with FL or ANN with GA etc. [7]. Other techniques
includes PSO-MPPT, ANN-MPPT, FL-MPPT and GA-MPPT
etc., for photovoltaic linked system working in mismatch
circumstances [8].
AI techniques are used to obtain Maximum Power Point
(MPPT) [4] such that maximum voltage and resultant
maximum current [5].Many shading mismatch conditions are
discussed and new topology namely DPP means Differential
Power Processing is suggested [6]. From results it has been
perceived that around 10%-20% drop in dc voltage of the
bridges in CHB can be expressed and AI techniques are
employed in order to predict the optimized value of firing
angles for bridges of CMLI.
The paper is structured as: Section I provides the introduction
of work. Section II defines the multilevel inverter particularly
CHB-MLI. Section III deliberates the modelling of PV
system.AI techniques such as ANN and FL and their working
are discussed in section IV. In section V, the implementation of
these techniques (ANN and FL) are done in PV fed CHB-MLI
system via look-up table. Simulation results, conclusion and
references are discussed and shown in the subsequent sections.
MULTILEVEL INVERTER
MLI offers several gains over a conventional two-level
inverter. It produce staircase output and as the number of levels
increase, output has more steps and therefore the harmonics in
the output voltage and therefore THD are minimized [9]. The
circuit diagram of n level CMLI is shown in Figure 1.
For n=7, 3 H-bridges are required and are connected in series
in order to obtain the desired output. The main benefit of
Cascaded H-Bridge MLI over others i.e. Diode Clamp and
Flying Capacitor is that it has simple switching and minimum
losses. Moreover, it is cheaper as it does not require extra
diodes and capacitors. Cascaded MLI is chiefly employed in
PV fed grid system as-
It provides reliability and energy whenever the PV panel
function under mismatch condition.
It performs transformer less action therefore cost effective.
The CMLI requires separate dc sources which can be found
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 13, Number 7 (2018) pp. 5057-5064
© Research India Publications. http://www.ripublication.com
5058
from batteries, solar cells or fuel cells.
A
S11
S41 S21
S31
A
S42 S22
S32
S1n
S4n S2n
S3n
S12
Vout
PV
PANEL 1V1
V2
Vn
PV
PANEL 2
PV
PANEL N
Bridge 1
Bridge 2
Bridge N
Figure 1. Single Phase Configuration of N Level MLI
In seven level Cascaded H-Bridge MLI, three H-bridges are
employed and voltage levels are from +3Vdc to -3Vdc and the
output voltage is the summation of the separate converter
output.
Voltage level (m) is given by -
m = 2s+1 (1)
where,
s- number of dc sources
Cascaded MLI are classified as – Symmetric and Asymmetric.
In symmetric CMLI values of DC sources are same whereas in
asymmetric type these values are unequal. Major benefit is that
it employs less switches. Moreover higher number of levels
are obtained and the levels increases proportionality in
comparison to the symmetrical type. The main fault in this
configuration is the loss of modularity such that the level upto
which the system constituents can be detached and associated
is lost. Acquired ac output voltage is the summation of output
derived from individual h-bridge.
The output voltage equation is given by-
V0 (θ) = ∑4𝑉𝑠
𝑛𝜋sin 𝑛𝜔𝑡∞
𝑛=1,3,5 (2)
THD equation is expressed as -
F(z) = % THD = √𝑉𝑜,𝑟𝑚𝑠
2 − 𝑉1,𝑟𝑚𝑠2
𝑉1,𝑟𝑚𝑠2 X 100 (3)
α1 α2 α3
+nVdc
-nVdc
0
π/2 3π/2π ωt
Figure 2. Output voltage waveform of seven level CMLI
MODELLING OF SOLAR PV SYSTEM
The solar array is formed by the combination of series and
parallel solar cells which provides the desired output voltage
and current under regular conditions. Terminal current of cell
is shown as the function of diode current, photo-generated and
shunt current. Figure 3 shows the equivalent circuit diagram of
PV Module.
Iph Id Ip
Rp
Rs
V
I
Figure 3. PV Module Equivalent Circuit
The photo-generated current (Iph) depends on temperature and
irradiance, calculated at standard conditions of irradiation Gref ,
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 13, Number 7 (2018) pp. 5057-5064
© Research India Publications. http://www.ripublication.com
5059
temperatureTc,ref and photocurrent Iph,ref as
Iph =G
Gref(Iph,ref + Iscc(Tc – Tc,ref)) (4)
Where,
Tc -actual working temperature of cell (K)
G - actual solar irradiance (W/m2)
Iscc-manufactured temperature coefficient of short-circuit
current (A/K)
Using the Shockley equation, diode current is -
Id = I0[expeVc
nKTc− 1] (5)
Where,
I0 - reverse saturation current (A)
Vc-voltage across diode (V)
e - electron charge (1.602x10-19 C)
η - quality or ideality factor
Rs- series resistance (Ω)
K - Boltzmann constant (1.38x10-23 J/k)
Reverse saturation current (I0) is given as-
I0 = I0,ref (Tc
Tc,ref)
3
exp ((eEg
nK) (
1
Tc,ref−
1
Tc)) (6)
Shunt current Ip is given as-
Ip = Vc
Rp (7)
where,
Rp- shunt resistance
The equation that depicts the I-V characteristics of PV
module is given by-
I = Iph – I0 (exp (e(V+IRs)
nKTc) − 1) −
V+IRs
Rp (8)
Figure 4. Grid Connected PV system
ARTIFICIAL NEURAL NETWORK IN CHB-MLI
The basic structural design of ANN is shown in Figure 5. It
comprises of three neuron layers namely - input, hidden, and
output layer. For feed-forward systems, the signal varies from
input to output via feedforward path. ANN is trained so that it
perform the task by modifying the values of the weights
amongst the elements. It is adjusted, on the basis of comparison
between output and target, unless the output of network meets
the desired target [10].
Figure 5. Architecture of Artificial Neural Network
In CMLI, each H bridge operate at different angle and the
output of CMLI is to be contingent on these angles. Usually,
these angles are computed off-line and put in the look-up table
for the functioning of CHB-MLI.
Figure 6. Block Diagram of Firing Angle Optimization System
For this work ‘Levenberg-Marquardt back propagation’
technique is used using NN Toolbox present in the MATLAB.
Using this, model of NN is made and is incorporated with the
seven level CHB inverter structure. Now the whole system is
employed in eliminating the harmonic in multilevel inverter.
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 13, Number 7 (2018) pp. 5057-5064
© Research India Publications. http://www.ripublication.com
5060
FUZZY LOGIC IN CHB-MLI
They have the advantage to be robust and relatively simple in
designing because they does not require the knowledge of the
exact model [11]. Fuzzy Logic (FL) method is built on the
fuzzy set theory and correlated method discovered by Lotfi
Zadeh. FL is a non-linear technique, tries to implement the
proficient knowledge of an expert in the designing of the
controller. Figure 7 shows block diagram of FLC. It comprises
of four components:
a) Fuzzification- It plans the crisp values into input fuzzy
sets in order to initiate the instructions or rules.
b) Rule Base-Instructions that describe the operation of
controller by employing a group of IF-THEN statements.
c) Inference Engine- It maps the input sets into the output by
applying the rules.
d) Defuzzification-It maps the output values of fuzzy into the
crisp values.
FU
ZZ
IFIC
AT
ION INTERFERNCE
ENGINE
DE
FU
ZZ
IFIC
AT
ION
RULE BASE
Input Output
Figure 7. Fuzzy Logic Interference System
GENETIC ALGORITHM
GA is one of the intelligence method and is built on the
mutative procedure. It is the presumptive, population based
optimization method, which is employed widely for
optimization purpose in various fields such as in engineering,
medical etc.
Figure 8. Flowchart of Genetic Algorithm
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 13, Number 7 (2018) pp. 5057-5064
© Research India Publications. http://www.ripublication.com
5061
In GA [12] prior to initialization the size of population and
iteration are fixed. Favorable solution of situation is termed as
chromosome. It is represented in binary or floating point
sequence.
𝑐ℎ𝑟𝑜𝑚𝑜𝑠𝑜𝑚𝑒 = (𝜃1, 𝜃2, 𝜃3) (9)
Now, conformity level of individual solution is determined then
on its basis individual is multiplied in subsequent iterations.
SIMULATION RESULTS AND DISCUSSION
The proposed seven level CHB-MLI structure is implemented
in solar fed system. As of look-up table, it is observed that
when H-bridges voltage are unequal then the angles
corresponding to it are predicted. A lookup table has been
made which comprises of diverse dc voltage with the voltage
deviation of about 20% means 5 V and resultant firing angles
acquired from the ANN and FLC.
A. Output obtain from proposed system by applying ANN technique:-
Figure 9. Case-I Output voltage waveform
Figure 10. Case-I Harmonic Spectrum
Figure 11. Case-II Output voltage waveform
Figure 12. Case-II Harmonic Spectrum
Figure 13. Case-III Output voltage waveform
Figure 14. Case-III Harmonic Spectrum
Figure 15. Case-IV Output voltage waveform
Figure 16. Case-IV Harmonic Spectrum
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 13, Number 7 (2018) pp. 5057-5064
© Research India Publications. http://www.ripublication.com
5062
Figure 17. Performance plot of ANN
B. Output obtain from proposed system by applying FL technique:-
Figure 18. Case-I Output voltage waveform
Figure 19. Case-I Harmonic Spectrum
Figure 20. Case-I Output voltage waveform
Figure 21. Case-II Harmonic Spectrum
Figure 22. Case-III Output voltage waveform
Figure 23. Case-III Harmonic Spectrum
Figure 24. Case-IV Output voltage waveform
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 13, Number 7 (2018) pp. 5057-5064
© Research India Publications. http://www.ripublication.com
5063
Figure 25. Case-IV Harmonic Spectrum
Figure 26. Control surface for the fuzzy model found by the GA
Table I. Lookup table of optimized switching angles and voltages
Cases Actual
Voltages (V)
Approximated
Voltages (V)
Firing angle obtained
from the lookup
table(0)
%THD
w.r.t
actual
values
(From
GA)
%THD w.r.t
approximated
values
(From ANN)
%THD w.r.t
approximated
values (From FL)
V1 V2 V3 V1 V2 V3 α1 α2 α3
I 88 89.5 90.5 90 90 95 100 200 280 14.39 11.39 9.59
II 90.2 92 93 95 95 95 60 170 100 14.02 11.58 9.63
III 95 93.4 96.6 95 95 100 90 150 300 13.69 11.91 9.64
IV 97.2 88.6 95.4 100 90 100 100 170 380 16.09 11.80 9.79
V 92.6 85 89.8 95 85 90 70 130 320 14.52 11.77 9.60
VI 99.8 89.7 96.3 100 90 100 90 190 390 15.59 11.63 9.61
Table II. Error Between GA-ANN and GA-FL
Cases % Error (GA-ANN) % Error (GA-FL)
I 3 4.8
II 2.44 4.39
III 1.78 4.05
IV 4.29 6.3
V 2.75 4.92
VI 3.96 5.98
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 13, Number 7 (2018) pp. 5057-5064
© Research India Publications. http://www.ripublication.com
5064
As shown in the table firing angle for PV connected 7 level
CHB-MLI system are optimized during actual as well as in
approximated condition. It has been observed that error
comes out to be lesser in case of GA optimized ANN system
as compared to GA optimized FL system. Therefore we can
say that GA optimized ANN system is superior in
performance from FLC system.
CONCLUSION
In this paper two AI techniques- ANN and FL are used and
Lookup Table employed angle optimization schemes are
implemented for PV fed seven level Cascaded H- Bridge
MLI system. These method are implemented and they gives
the results of lower harmonics and of higher accuracy. All
these results are justified in MATLAB by taking voltage
variation of about 20%. It is detected that the techniques
worked suitably for intermediate values of voltage and as
well as with very less difference in the harmonic. Whenever
there is a small difference in approximate voltages and
actual voltage, ANN and FL- using GA optimized lookup
table methodology can be applied. From results, it can be
seen that error is lesser in case of GA-ANN than that of GA-
FL. Further these techniques are extended to hybrid system
like ANFIS that is a combination of ANN and FL. The work
done in this paper is first step prior to field installation in
order to experimentally authorise the effectiveness of the
proposed system.
REFERENCES
[1] Balamurugan M., Sarat Kumar Sahoo, Sukruedee
Sukchai, “Application of soft computing methods for
grid connected PV system: A technological and status
review”, Renewable and Sustainable Energy
Reviews, Volume 75, August 2017, Pages 1493-
1508.
[2] N. Ammasai Gounden, Sabitha Ann Peter, H.
Nallandula, S. Krithiga, “Fuzzy logic controller with
MPPT using line-commutated inverter for three-
phase grid-connected photovoltaic systems”,
Renewable Energy, Volume 34, Issue 3, March 2009,
Pages 909-915.
[3] Yi-Hwa Liu, Shyh-Ching Huang, Jia-Wei Huang,
Wen-Cheng Liang, “A Particle Swarm Optimization-
Based Maximum Power Point Tracking Algorithm
for PV Systems Operating Under Partially Shaded
Conditions”, in Energy Conversion, IEEE
Transactions on , vol.27, no.4, pp.1027-1035, Dec.
2012.
[4] A. Mellit, S.A. Kalogirou, L. Hontoria, S. Shaari,
“Artificial intelligence techniques for sizing
photovoltaic systems: A review”, Renewable and
Sustainable Energy Reviews, Volume 13, Issue 2,
February 2009, Pages 406-419.
[5] Hong Hee Lee, Le Minh Phuong, Phan Quoc Dzung,
Nguyen Truong Dan Vu, Le Dinh Khoa, “The new
maximum power point tracking algorithm using
ANN-based solar PV system”, TENCON 2010 - 2010
IEEE Region 10 Conference , vol., no., pp.2179-
2184, 21-24 Nov. 2010.
[6] Olalla, C.; Deline, C.; Clement, D.; Levron, Y.;
Rodriguez, M.; Maksimovic, D., “Performance of
Power-Limited Differential Power Processing
Architectures in Mismatched PV Systems,” in Power
Electronics ,IEEE Transaction on,vol.30, no.2,
pp.618-631, Feb. 2015.
[7] Anil K. Rai, N.D. Kaushika, Bhupal Singh, Niti
Agarwal, “Simulation model of ANN based
maximum power point tracking controller for solar
PV system”, Solar Energy Materials and Solar Cells,
Volume 95, Issue 2, February 2011, Pages 773-778.
[8] T. Esram and P. L.Chapman, “Comparison of
photovoltaic array maximum power point tracking
techniques”, IEEE Trans. Energy Convers., vol. 22,
no. 2, pp. 439–449, Jun. 2007.
[9] P. Shanmuga Aravind and S. Albert Alexander,
“Harmonic Minimization of a Solar fed Cascaded H
Bridge Inverter using Artificial Neural Network”,
International Conference on Energy Efficient
Technologies for Sustainability (ICEETS), pp. 163 –
167, April 2013.
[10] Jagdish Kumar, Jaimala Gambhir and Anil Kumar,
“Control of Switching Angles for a CMLI using
ANN”, Proceedings of 2014 RAECS UIET Panjab
University Chandigarh, 06 – 08 March, 2014.
[11] A. Messai, A. Mellit, A. Guessoum, S.A. Kalogirou,
“Maximum power point tracking using a GA
optimized fuzzy logic controller and its FPGA
implementation”, Solar Energy, Volume 85, Issue 2,
February 2011, Pages 265-277.
[12] Ozpineci B, Tolbert L. M, and Chiasson J. N. (2005),
“Harmonic optimization of multilevel converters
using genetic algorithms”, IEEE Power Electronics
Lett., Vol. 3, No. 3, pp. 92–95,2005