You are on page 1of 6

2017 International Seminar on Application for Technology of Information and Communication (iSemantic)

Analyze and Optimization of Genetic Algorithm


Implemented on Maximum Power Point Tracking
Technique for PV System

Prisma Megantoro Fransisco Danang Wijaya


Department of Electrical Engineering and Information Department of Electrical Engineering and Information
Technology Technology
Universitas Gadjah Mada Universitas Gadjah Mada
Yogyakarta, Indonesia Yogyakarta, Indonesia
prisma.megantoro@mail.ugm.ac.id danangwiyaya@ugm.ac.id

Eka Firmansyah
Department of Electrical Engineering and Information Technology
Universitas Gadjah Mada
Yogyakarta, Indonesia
eka.firmansyah@ugm.ac.id

Abstract— In photovoltaic system, Solar Charge Controller energy nowadays which has energy power for unlimited
(SCC) device has vital role to keep the performance to get system sunlight. The application of this kind energy is also
efficiency higher. It is done by tracking the maximum power developed followed by converter devices technology
point on solar cell array where power transfer is on its maximum development. Photovoltaic has system used to provide
condition. This level is on V-I curve where curve characteristic electrical energy as the source of it. One of vital device on
depends on temperature and sun irradiation level happen on the this system is Solar Charge Controller that has function to
moment. There are a lot of algorithms have been developed to optimize power transfer from electrical power to electrical
increase the efficiency to this process, it is included to artificial storage, or from solar panel to battery bank. To get
intelligence based algorithm. One of them is genetic algorithm as
maximum power transfer, it needs to do maximum-power-
heuristic method developed to imitate the natural selection
point-tracking (MPPT) by converting solar-panel-output-
process. The implementation of this algorithm is conducted for
tracking process of maximum power point. This research used voltage to maximum-power-point-voltage (Vmpp).
modelling to compare this tracking technique to tracking process Some methods have been developed to increase the
which used analytical method. Furthermore, correlation analysis performance of MPPT on SCC device [3][4]. More efficient
was done to parameters used to output parameter. Research system has more accurate and faster performance. On its
results showed that the implementation of genetic algorithm to technology development, adopted method from artificial
maximum power point had excellently on tracking accuracy intelligence based algorithm has tried to change the
parameter and tracking time. While correlation test showed that
conventional method [5]. There are many artificial
parameter of individual number and generation number effected
significantly to the increasing of tracking accuracy and tracking
intelligences for this application toward evolutionary
time. algorithm like in this research [6][7][8]. This research
implements genetic algorithm used to modelling MPPT
Keywords— maximum power point tracking; genetic which is considered to photovoltaic characteristics, V-I
algorithm; photovoltaic; curve. Performance analysis from its implementation was
conducted by comparing the existing method performance
[9] with comprehensive counting. Different genetic
I. INTRODUCTION algorithm has random-voltage-points. This paper aims to do
Nowadays, renewable energy enters new development parameter analysis which effect to algorithm performance,
era that significantly happens and it is followed by the correlation test was also done to sampling with some
finding of unlimited energy resources and technology [1]. iterations on each variation from each tested parameter to
Used technology strives for changing to be more advance to obtain the significant parameters effect to performance of
fulfill higher demand of electricity [2]. More efficient of genetic algorithm implementation in tracking the maximum-
renewable energy using, bigger number of energy that is power-point.
used for electricity. Photovoltaic is a popular renewable

79
2017 International Seminar on Application for Technology of Information and Communication (iSemantic)

II. METHOD
is constanta of energy gap,
MPPT technique by using genetic algorithm need to have
the modelling first to obtain the data curve.

A. Photovoltaic Modelling (5)


Solar panel is modelled without any connection to
battery, and neglect the partial shading effect. It is assumed Modelling result shows different V-I curves which
that solar panel has full lighting. Characteristic of the solar changed depend on level of sun irradiation on full lighting
panel is modelled by V-I curve which shows the relationship as shown in Figure 3[12].
between voltage and current of solar panel as shown in
Figure 1 [10].

Fig.1. V-I Curve to solar panel on full lighting


Fig.3. V-I curve on its irradiation level variation
P-I curve also shows the relationship between voltage
and solar panel power. The effected factor to both
Figure 4 showed the different of V-P curve [12].
characteristics are temperature and sun irradiation level[11].
The model of solar panel module consists of some parallel
PV cells. The characteristics are represented on Figure 2 and
equation (1)[12].

Fig. 4. V-P curve on its variation of irradiation level

There were 4 types of solar panel as a sample, each solar


Fig. 2. The series of solar panel model panel was tested by each specification.
TABLE 1. SPECIFICATION OF SOLAR PANEL TYPE
ܸ ൅ ‫ܫ‬Ǥ ܴ‫ݏ‬ ܸ ൅ ‫ܫ‬Ǥ ܴ‫ݏ‬
‫ ܫ‬ൌ ‫ݒ݌ܫ‬ െ ‫ ݋ܫ‬൤݁‫ ݌ݔ‬൬ ൰ െ ͳ൨ െ
ܽǤ ܸܶ ܴ‫݌‬ Parameter Type of Module
(1) (unit)
KC85T KD135SX MSX60 M2453BB
a is ideal diode constanta, is thermal junction Iscn (A) 5.34 8.37 3.8 8.7
voltage, with is defined by Vocn (V) 21.7 22.1 21.1 37.7
Imp (A) 5.02 7.63 3.5 8.2

(2) Vmp (V) 17.4 17.7 17.1 30.1


is number of PV cell which is connected in series, k Ns (unit) 36 36 36 60
is Boltzmann (J/K)constanta, T is temperature (F) ambient,
while q is electron charge (C).
Iscn is short-connected-current measured on one series
connected of PV cell, Vocn is open-connected-voltage one
(3) series connected of PV cell, while Imp and Vmp are current
and maximum-power-point-voltage gained from laboratory
G is sun irradiation level (W/m^2), is scaled testing on each factory or producer. Ns is number of
current/temperature coefficient (A/K). PV cell on one series of solar panel module.
V-I dan V-P curves of the model sample are shown in
Figure 5 and Figure 6.
(4)

80
2017 International Seminar on Application for Technology of Information and Communication (iSemantic)

those two processes, the new individuals are born and


will join to prime individual to form new population on
the next generation. However, in this algorithm
modelling, it has limitations in population size and
individual numbers. Here, the population size is fixed.
The next generation population is formed by selecting
prime chromosome and baby chromosome based on
fitness function or it is called strength level on natural
process.
The mapping of natural process into computation
process is shown on Table 2.
TABLE 2. THE MAPPING OF NATURAL PROCESS INTO
COMPUTATION PROCESS

Fig. 5. V-I curve on module type on different solar panel Natural process Computation process
Individual Problem solving
Population Solving collection
Value of fitness Solving quality
Chromosome Solving representation
Gen Part of solving representation
Natural selection Selecting the problem solving based on
its quality
Crossing Genetic operator
Mutation Genetic operator

b. The implementation of GA-MPPT


The process of maximum-power-point-tracking is
explained on current chart below as shown in Figure 7.

Fig. 6. V-P curve of module type on different solar panel

The difference on each tested solar panel specification


has shown different curve of V-I and V-P.

B. Genetic Algorithm-MPPT Modelling


a. Theory of genetic algorithm
Genetic algorithm (GA) is a heuristic method based
on genetic principle and natural selection process of
theory of Darwinian Evolution. This optimization
method was developed by John Holland around 1960s
and is popularized by his student, David Goldberg
around 1980s [13].
On Theory of Darwinian Evolution, individual is
created randomly then exist by reproduction so that
group of individual is exist, namely population. This
theory recognizes natural selection process to keep
individual by only its high strength to be exist in life. It
also works on genetic algorithm. The optimization is
started by generating individual randomly, which is
called as chromosome. Those chromosomes consist of
some gens which are represented to binary number n-
bit. Their real value will be observed to represent their
strength level. Only chromosome with the highest
strength that can be exist.
Chromosomes are also crossed like the process of
creature copulation. Some chosen gens are crossed
between prime chromosome with other chromosome.
Then mutation happens to these chromosomes. From Fig. 7. Flowchart of GA-MPPT algorithm

81
2017 International Seminar on Application for Technology of Information and Communication (iSemantic)

c. Declaration and parameter initialization chromosomes. This crossing is using n-point method.
Before MPPT by GA begins, declaration from The crossing point is selected randomly. Number of
parameters used on solar panel characteristics crossed gen is considered from its controlled crossing
modelling is done. Those declared parameters are probability.
ambient of temperature value and level of sun
irradiation. (7)
Initialization of free algorithm parameter is also done to
define performance of tracking process. The parameters In this method, every couple of chromosome is cut into
are number of gen, number of individual, number of 2 pieces. The first baby chromosome is coming from
generation, crossing probability, and mutation the first cut of second prime chromosome. And the
probability. second baby chromosome is coming from second cut of
The management of algorithm parameter value as first prime chromosome and second cut of second prime
suggested by [14] are follows: chromosome.
a. Quite big crossing probability (60-70%) i. Mutation
b. Quite small mutation probability (1 gen for 1 Some genes on each chromosome will be mutated.
chromosome/individual) The choosing of gen number is conducted randomly
c. Population size between 50-500 while mutated gen number is considered by its mutation
chromosome/individual probability.
d. Generating of first prime chromosome
Random matrix generating is conducted to (8)
represent gen from whole generated individual. Matrix
is generated by size (number of gen x number of Mutation method used binary code. The chosen gen
individual). Every element of matrix is filled by 0 or 1. will be converted into its logical condition. If the
Herewith, each chromosome is represented by a binary beginning is 1 then being 0, and vice versa. Then the
code gen-bit number. mutated chromosome will be used as prime
e. Gen coding chromosome for the next generation.
Every represented chromosome in binary code is Optimization process done in genetic algorithm will
converted into real number to define its representation stop after any requirement is achieved [15]. For that
value on equation: case, the requirement is number of generation.

(6) III. IMPLEMENTATION


x is chromosome representation value, is
A. GA-MPPT Performance
minimum value (0), is maximum voltage value a. Comparison testing
(Voc), N is number of binary bit/ gen, is binary MPPT tracking performance uses genetic algorithm
number. The value of x represents the value of solar will be compared to MPPT done by [10] which uses
panel voltage as controlled parameter. comprehensive counting by same equations. This is
f. Evaluation done for different 4 variations of solar panel module
Every representation value on every chromosome type. Every testing has 33 same iterations to genetic
then added into equation (Villalva) modelling to obtain algorithm parameters. The treatment is done to number
its fitness function value, which is value of solar panel of gen (12 times), number of individual (12 times), and
power (P). number of generation (40 times). Those tests are
g. Selection considered from some main parameters.
By using level selection method and elitist
selection method, every chromosome will be ordered
according to its fitness value. Chromosome with higher
fitness value will be put on the front order. The
probability on chosen chromosome is defined based on
linear distribution function. Then, those chromosomes
on half-order from its individual number (and up) will
be kept for the next generation. It means that remain
chromosomes are omitted.
h. Crossing
Selected chromosomes then being crossed by other
chromosomes. Each chromosome has its couple. Every
couple of prime chromosome will produce 2 baby

82
2017 International Seminar on Application for Technology of Information and Communication (iSemantic)

TABLE 3. THE RESULT OF TRACKING PERFORMANCE TESTING ON COMPREHENSIVE ALGORITHM AND GENETIC ALGORITHM
Type of PV KD135SX_UPU M2453BB MSX60 KC85T

Algorithm Villalva GA-MPPT Villalva GA-MPPT Villalva GA-MPPT Villalva GA-Mgn


Pmpp (P) 135.0432 135.5347 246.8196 231.0215 59.8485 58.1707 87.3477 84.9592
Vmpp (V) 17.901 17.4846 30.16 29.1431 17.302 16.7318 17.577 17.562
Impp (A) 7.5439 7.7516 8.1837 7.9272 3.4591 3.5364 4.9694 4.8377
Accuracy (%) 98.8644 98.7833 99.807 96.8202 98.8187 97.8568 98.98.28 99.069
Period (second) 0.63801 0.22088 1.0972 0.10457 0.98735 0.20485 0.91242 0.20163
Voc (V) 22.1 37.7 21.1 21.7
Interval of voltage (V) 0.221 0.005 0.377 0.009 0.211 0.005 0.217 0.005

According to the monitoring data of PV System


built in The Kinanthi Hostel of Universitas Gadjah
Mada, captured at Friday May 26th 2017, the output
power is significantly fluctuated. The trend of the
output power shows the irradiance level that
proportionally also fluctuated. The fluctuation indicates
that the sky condition changes slowly that was obtained
11-13 times per hour. Based on the value of tracking
period showed on Table 3, the GA-MPPT algorithm
will not give a significant effect to the performance of
the SCC device by its tracking period.
The figures 8, 9, 10 and 11 showed the comparison
of these two algorithms based on the V-P curve. Beside Fig. 10. V-P curve on comparison testing by using solar panel
shows the output power, they also shows the accuracy. module type MSX60
Nearer to the blue spot (Vmpp) from PV module
specification), mean it has higher accuracy.

Fig. 11. V-P curve on comparison testing by using solar panel


Fig. 8. V-P curve on comparison testing by using solar panel module type module KC85T
KD135SX UPU
b. Correlation testing
The performance of maximum power point
tracking by using genetic algorithm is effected by the
value of used algorithm, so that correlation testing to
each input and output parameter is done. Every testing
needs 33 iterations. The result can be seen in Table 4.

Fig. 9. V-P curve on comparison testing by using solar panel


module M2453BB

83
2017 International Seminar on Application for Technology of Information and Communication (iSemantic)

TABLE 4. THE RESULT OF CORRELATION TESTING BETWEEN REFERENCES


INPUT PARAMETER AND OUTPUT PARAMETER OF MAXIMUM
POWER POINT TRACKING [1] B. Parida, S. Iniyan, and R. Goic, “A review of solar photovoltaic
Output\Input Number of Number of Number of technologies,” Renew. Sustain. Energy Rev., vol. 15, no. 3, pp. 1625–
Gen Individual Generation 1636, 2011.
Pmpp 0.7801 0.807 0.8337 [2] J. Khan and M. H. Arsalan, “Solar power technologies for sustainable
electricity generation – A review,” Renew. Sustain. Energy Rev., vol.
Time 0.2843 0.9838 0.9989 55, pp. 414–425, 2016.
Vmpp Accuracy 0.7506 0.9313 0.8858 [3] D. Verma, S. Nema, A. M. Shandilya, and S. K. Dash, “Maximum
power point tracking ( MPPT ) techniquesௗ: Recapitulation in solar
photovoltaic systems,” Renew. Sustain. Energy Rev., vol. 54, pp.
1018–1034, 2016.
IV. CONCLUSION [4] K. Ishaque and Z. Salam, “A review of maximum power point
tracking techniques of PV system for uniform insolation and partial
This research has some conclusions regarding to shading condition,” Renew. Sustain. Energy Rev., vol. 19, pp. 475–
performance of MPPT by using genetic algorithm as follows: 488, 2013.
[5] B. Bendib, H. Belmili, and F. Krim, “A survey of the most used
1. Characteristics modelling of solar panel module has MPPT methods: Conventional and advanced algorithms applied for
successfully done with V-I and V-P result and those photovoltaic systems,” Renew. Sustain. Energy Rev., vol. 45, pp.
are effected by specification of solar panel module 637–648, 2015.
[6] N. Hashim, Z. Salam, and S. M. Ayob, “Maximum Power Point
2. The modelling of maximum power point tracking Tracking for Stand-Alone Photovoltaic System Using Evolutionary
has successfully done by using genetic algorithm Programming,” no. March, pp. 7–12, 2014.
[7] M. F. N. Tajuddin, S. M. Ayob, Z. Salam, and M. S. Saad,
3. The performance of maximum power point tracking “Evolutionary based maximum power point tracking technique using
from output power and tracking accuracy point of differential evolution algorithm,” Energy Build., vol. 67, pp. 245–
view has found that Villalva has better quality. But 252, 2013.
from processing time point of view, GA method has [8] A. soufyane Benyoucef, A. Chouder, K. Kara, S. Silvestre, and O. A.
better quality. This happens because Villalva sahed, “Artificial bee colony based algorithm for maximum power
method did comprehensive counting while GA point tracking (MPPT) for PV systems operating under partial shaded
conditions,” Appl. Soft Comput., vol. 32, pp. 38–48, 2015.
method only counted the chosen voltage randomly,
that is prime chromosome. [9] M. G. Villalva, J. R. Gazoli, and E. R. Filho, “Comprehensive
Approach to Modeling and Simulation of Photovoltaic Arrays,” vol.
4. The tracking performance is effected by algorithm 24, no. 5, pp. 1198–1208, 2009.
parameter: used number of gen has no effect [10] M. G. Villalva, J. R. Gazoli, and E. R. Filho, “Modeling and Circuit-
Based Simulation of Photovoltaic Arrays,” vol. 14, no. 1, pp. 1244–
significantly to output power (Pmpp) and tracking 1254, 2009.
accuracy (Vmpp), even to tracking time. But
[11] P. Kumar, G. Jain, and D. K. Palwalia, “Genetic Algorithm Based
number of individual is the most significant Maximum Power Tracking in Solar Power Generation,” pp. 1–6,
effected parameter to tracking accuracy. While 2015.
number of generation is the most significant [12] M. Seyedmahmoudian and R. Rahmani, “Simulation and Hardware
effected parameter to tracking time. Implementation of New Maximum Power Point Tracking Technique
for Partially Shaded PV System Using Hybrid DEPSO Method,”
IEEE Trans. Sustain. Energy, vol. 6, no. 99, pp. 1–13, 2015.
[13] Haupt, R.L. dan Haupt. “Practical Genetic Algorithm”, New Yersey:
John Wiley and Sons, Inc.,2004.
[14] Hopgood, A.A, “Intelligent Systems for Engineer and Scientists”,
Boca Raton: CRC Press LLC., 2001.
[15] Ding, Q and Gasvoda, “A Genetic Algorithm for Clustering on Image
Data”, International Journal of Computational Intelligence 1(1):81-
86,2004.

84

You might also like