Genetic Algorithm Based Maximum Power

Tracking in Solar Power Generation

Prakash Kumar, Gaurav Jain, Dheeraj Kumar Palwalia
UCE, Rajasthan Technical University, Kota, Rajasthan, India

study of GA and PO for exploration and exploitation of

Abstract— This paper presents genetic algorithm (GA) based renewable energy sources optimally.
maximum power point tracking (MPPT) for photovoltaic (PV)
array integrated with battery storage unit (BSU) as power
II. MODEL OF HYBRID PV-BSU SYSTEM
generation unit in standalone mode. PV generation depends on
solar irradiance, site location and environmental factors like Dynamic PV hybrid model consisting of individual PV array
temperature, wind and. Thus output of PV output is fluctuating in and BSU is connected to point of common coupling (PCC)
nature and the addition of non linear load make the situation through bus bars. It is assumed that PV output power changes
more critical. GA based MPPT for PV generation works for local
round the clock. Partial shading condition has been neglected,
optimal solution. DC/DC and boost converter has been used to
obtain rated desired rated voltage. BSU acts as secondary source i.e. PV array is either fully exposed to sun light or fully shaded,
to ensure uninterrupted power supply. A suitable three phase to avoid reverse biasing condition or local heat spot formation
voltage source inverter (VSI), at point of common coupling (PCC), in PV module [10]. The modern PV module consists of PV
provides regulated power supply. Boost converters incorporating cells, mounting structure, MPPT mechanism, converters,
proportional controller (PI) increases PV voltage up to 400V. storage unit, electrical & mechanical connections to regulate
Conventionally used perturb and observe (PO) algorithm has and utilize electrical output [11]. The electrical power output of
been used for comparative study. Controlled pulse width PV module depends on electrical, thermal, PV spectral and
modulated (PWM) switching pulse for converter unit and single optical property of PV cell array and PV angle & irradiance
section LC filter at output terminal eliminates harmonic
[12]. The technical problems such as islanding detection,
components. To exhibit the system compatibility resistive,
reactive, asynchronous induction motor and nonlinear load has harmonic distortion, electromagnetic interference and low
been switched at different instants for variable solar irradiance efficiency of PV cells are major bottle neck for widespread
and temperature. application of PV systems [13].
A. PV Power Generation
Index Terms— battery storage, fuel cell, MPPT, genetic
algorithm, PV generation, voltage source converter. Conventionally PV cells were designed based on band gap
energy (eV). Low band gap energy have high current (I=eNA)
I. INTRODUCTION but low voltage (V=Eg/e), and vice versa. Here, e is electron
The high solar potential, i.e. about 20,000 times the total charge, N is the number of photons, Eg is energy gap, and A is
world demand of electricity, has not yet been exploited due to the surface area of solar cell. Therefore, it was preferred to use
technical and economical lags. Solar energy is omnipresent and materials with energy gap between 1 and 1.8 eV like crystal
available round the globe free of cost. Beside all these silicon (1.12 eV), amorphous silicon (1.75 eV), copper indium
advantages, PV generation is sure to remain unavailable during
diselenide (1.05eV), cadmium telluride (1.45 eV), gallium
night time. Therefore, secondary power source like BSU need
to be integrated with PV to assure uninterrupted power supply arsenide (1.42 eV) and indium phosphate (1.34 eV) [14, 15].
in standalone mode. BSU also acts as power conditioner and These modules are rated in terms of peak kilowatts (KWp) i.e.,
integrated VSI minimizes THD as well as works as active the amount of expected electrical power output when the sun is
power filter. In order to track maximum power a number of directly overhead on a clear day.
approaches, commonly known as maximum power point A solar cell is generally represented by one diode model as
tracker (MPPT), have been made. Among the commonly
shown in fig. 1.
applied MPPT approaches; like purterb & observe (PO), power
IPV
matching, incremental conductance, fractional open-circuit
voltage/short-circuit current, power differential feedback IC RS
ID IR P
control, curve fitting, dc-link capacitor droop control,
+
intelligent control, and some other special control methods; PO Dependent
Parameters
Constant
(Temp,
Current
VD RP VPV
is commonly used due to its simplicity and good results [1-9]. Irrediance, etc.
Generator

-
Due to dynamic nature of GA, it has drawn attention of
power system researchers to explore and exploit renewable
energy sources optimally. This paper deals with comparative
Fig: 1 One diode model of PV cell

978-1-4799-8371-1/15/$31.00 2015
c IEEE 1
 The model includes a constant current sourcce representing RI
I sc = I PV − I 0 ª¬e( s sc
nkT )
− 1¼º − ¨§ s sc ·
eR I
PV cell, a diode, a shunt resistance RP and a sseries resistance
© Rsh ¸¹
RS. The basic terms involved in PV conversioon are given as: (7)
V .I Pm
VD And, FF = m m =
I PV = I C − I D − = I C − I D − I RP Voc .I sc Voc .I sc
RP (1)
(8)

VPV = VD − Rs I PV (2)
Here, Si is solar irradiance, Voc is open circuit voltage, Vsc is
short circuit voltage, and FF is fill faactor (0.5 to 1).

(
I D = I 0 e(VD nkT )
−1 ) (3)
§ NOCT − 20 · 0
T = Ta + ¨ ¸ Si
© 0.8 ¹ (4)
Where, I0 is saturation current, VD is dioode voltage, ID is
diode current, k is Boltzmann’s constant (1.338x10-23J/K), n is
numerical constant (1 for germanium and 2 for silicon), T is
junction temperature, Ta is ambient temperatuure, and NOCT is
cell temperature in a module when ambient ttemperature is 20
0
C. Fig: 4 Change in the characteristtics due to the change in
solar insolations

B. Battery Storage Unit

Due to simplicity and robust workinng if lead acid battery; it is
generally used with PV module. Th he equivalent circuit of the
battery is shown in fig 5. The param
meters of battery have been
given in appendix.

Fig:2 Change in the characteristics due to tthe change in

temperature

Fig. 5 Equivalent circuitt of the battery

III. OBJECTIVE FUNCTIO

ON FORMULATION
Fig: 3 Typical I-V characteristic of a solar cell in
The constraints taken under consiideration include power of
steady-state operation
the PV module and battery. GA and PO aim at tracking
Conversion efficiency (Ș) of PV cell is givven as the ratio of maximum power output and minimiizing the objective function
maximum power output (Pmo) and input poweer (Pin) at a given to global optimal result. Algorithm traces global maxima and
cell temperature T. minima by voltage versus poweer curve. Power balance
constraints for any period includees power supplied by the
Pmo Pmo (Voc .I sc ) FF hybrid generation and load demand,, i.e. generated power must
η= = = meet total load demand, satisfying power quality and reliability
Pin ASi ASi (5) constraints.
nkT ª § I L + I 0 · º
Where, Voc = « log ¨ ¸» PPV + PBSU ≥ Pload
e ¬ © I0 ¹¼ (9)
(6)

2 2015 International Conference on Power and Advanced Control Engineering (ICPACE)

 From power curve of selected PV module, it is obtained that 2. Once a bee becomes an EB, it remains EB until FS is
voltage increases from 0 to VPV. Power curve first increases to a depleted temporarily or permanently.
maximum point and then gradually decreases to minimum 3. UB always look for FS by watching waggle dance,
value. PV module operating condition is selected so as to assure employ themselves by motivation or become SB.
maximum power operating point. Partial shading condition has 4. Each FS has dedicated EB, SB and common OB.
been neglected to overcome malfunctioning conditions of PV 5. EB always updates information with OB and SB
array. regularly investigates abandoned FS.
In order to track MPPT with ABC, gbest-ABC, and PO; Algorithm starts with no previous information about best FS.
transformation of Y-axis has been suggested. Axis First of all variables like total population (TP) of bee, maximum
transformation is done so as to obtain mirror image of voltage number of cycles (MNC), objective function, and function
power graph. It is given as: variables are defined. Function for calculation of fitness of
objective function for each FS is initiated by minimizing the
Pnew = (PPV,max + 50 ) − Pinst
(10)
greedy selection function. Depending on information of quality
of nectar updated by EB and hive storage, nectar is carried by
And, N PV , N BSU ≥ 0 EB to the hive. Excess nectar is stored separately for future
(11)
usages.
Here Pnew is new transformed value, Ppv max is maximum Food position gets updated in every cycle as per fitness
power point in power curve, and Pinst is instantaneous power. solution by EB and SB. For known and abandoned FS, position
The factor 50 is taken to avoid interference in transformation at of each FS is represented in three dimensions viz. population (i
Ppv,max and obtain effective transformation. = 1, 2, ..., TP), dimensional variable (j= 1, 2, ..., D), and specific
For BSU, it is assured that objective function (k= 1, 2, ..., OF). Number of cycle represents
number of iterations and needs to be minimized for optimal
PBSU,min ≥ PBSU ≥ PBSU,max solution in a given position.
(12)
And, PBSU,min = (1 − DOD ) .PBSU,max If the upper and lower bounds of ith bee xi for kth variable in jth
(13) direction is xminjk and xmaxjk respectively and rand [0, 1] is a
uniformly distributed random number in the range [0, 1]; SB
Here, PBSU,min, PBSU, and PBSU,max are minimum, actual and initiates search of new FS as:
maximum capacity of BSU respectively; and DOD is depth of
discharge of BSU.
(
x ijk = x min jk + rand [ 0,1] x max jk − x min jk ) 14)
IV. GENETIC ALGORITHM
A. Gbest-ABC Algorithm When SB finds a food source, it becomes EB and returns to
hive with nectar and updates food position xijk with OB. In the
GBest ABC is based on GA problem based on behavior of consequent cycles, OB checks new position with previous
honey bee. It incorporates honey bees into employed and position xpjk and updates the information as:
unemployed bees (UB). Number of employed bees (EB) is
equal to UB. 10 to 15% of UB are scout bees (SB) and rests are
onlooker bees (OB). SB search new and abandoned food source (
vijk = x ijk + φijk x ijk − x pkj ) (15)
(FS) randomly without prior knowledge and gather the
information about the quality of FS. As soon as SB finds new Where, i, j are randomly chosen and ij. ijk is a random
FS, it becomes EB and carries nectar to hive. It deposits nectar number between [-1, 1]. After generating new position, optimal
at hive and starts waggle dance to motivate OB. Waggle dance position is updated and old one is removed from memory. The
is one of the important activities because updated food probability pi that the new FS position is optimal depends on
information is shared in the dancing area. OB stays in the hive, solution fitness solution given by:
watches waggle dance and updates food quality information. ( fitness )i
Food quality generally depends on distance of FS from hive, pi = TP
amount of nectar and food quantity. OB always have option of
either becoming SB, stay at hive to watch waggle dance and
¦ ( fitness )n (16)
n =1
update information gathered by EB or become EB by
motivation of one of the best options among many dancing In order to achieve better feasible solution, FS must be
bees. When EB return to hive with nectar, they also have choice explored and exploited optimally. ABC algorithm is good at
to remain an employed bee, waggle dance and motivate OB to exploration but not in terms of exploration as the probability of
exploit FS or become OB. This process continues until food getting a good FS and a bad FS is equal. So by getting inspired
source is depleted and all employed bees become unemployed. by other algorithms like particle swarm optimization,
exploitation capability of gbest ABC can be improved as:
Assumptions made:
1. Food quality depends on quality & quantity of nectar ( )
vijk = x ijk + φijk x ij − x pjk + ψ ijk (y jk − x ijk )
and distance of FS.
(17)

2015 International Conference on Power and Advanced Control Engineering (ICPACE) 3

 Where, Ȍijk is a non-negative random number in range [0, C],
yjk is jth term of global best solution. Parameterr “C” controls the
rate of exploitation and exploration. Higher vvalue of “C” leads
to better exploration and lower exploitation; hence need to be
taken carefully.
B. Perturb and observe Algorithm
In this algorithm a slight perturbation is iintroduced in the
system. Due to this perturbation, module pow wer changes. If the
power increases due to the perturbation then tthe perturbation is
continued in that direction. After the peak pow
wer is reached the
power at the next instant decreases and hennce after that the Fig. 8 MPPT buck converteer output voltage
perturbation reverses.
The PV array output current & voltage are shown in fig. 7.
When the steady state is reached the alggorithm oscillates
The PV array Current is 5A and voltaage is 45 volts. MPPT buck
around the peak point. In order to keep thee power variation converter result has been shown in fig.
fi 8. It is observed that the
small the perturbation size is kept very small.. The algorithm is MPPT buck converter maintains th he PV array output to 45
developed in such a manner that it sets a refference voltage of
volts. The output of boost converteer is depicted in fig 9. The
the module corresponding to the peak voltage of the module. A
voltage is boosted to 400 volts.
PI controller then acts moving the operatiing point of the
BSU is used as back-up or storrage purpose. Discharging
module to that particular voltage level. It is obbserved that some
characteristic is shown in fig 12. BSU used here contains
power loss occur due to this perturbation. Alsoo the method has a lead-acid battery due to long life an nd simplicity. The DC link
disadvantage to track the power under fast varrying atmospheric voltage is shown in fig. 11. It is also
o around 400 volts.
conditions. Yet this algorithm is popular and considered
simple. Output line voltage and current waveforms
w has been shown
in fig 10. In fig 10, it is observed thaat the designed filter works
satisfactorily and the output wave fo orms are sinusoidal.

500

450

400

350
Boost converter2 voltage

300

Fig. 6 Perturb and observe algorith

hm 250

V. SIMULATION RESULTS 200

150

Simulation has been done in matlab simuulink platform to

100
obtain results of hybrid PV-BSU generatiion system. The
simulation results for PV array, MPPT buckk converter, boost 50

converter, 3-phase inverter and LC filter has bbeen obtained. 0

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2
Time

6 Fig. 9 Boost converter output

o voltage
5
PVCurrent

0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0
Time

50

40
PVVoltage

30

20

10

0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0
Time

Fig: 7 PV array current and voltage for P

PVA alone Fig. 10 Line voltage and currrents wave forms

4 2015 International Conference on Power and Advanced Control Engineering (ICPACE)

 500
Magnitude of three line voltages for
fo different values of L and
Dc Link Voltage 400 C has been given in table 2. Frrom table it is seen that
fundamental values of three line voltages
v are maximum at
300 L=1.63 mH, and C=450ȝF.
200 Fig. 13 shows comparative curve for
f MPPT techniques using
ABC, gbest ABC, and PO using pow wer curve for PV irradiance.
100
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 PO is the most robust and simplee in application but lacks
Time exploitation of PV power at high her irradiance. At normal
Fig. 11 DC link voltage irradiance below 650 W/m2, PO has been
b found to exhibit better
results while for higher irradiance,, gbest ABC gives higher
power output. Table 3 shows power obtained at different
irradiance value.

Fig. 12 Battery discharge chaaracteristic

THD obtained for all three line voltage wavveforms has been
given in table 1. THD is lowest for the inducttor ‘L= 1.63 mH’ Fig. 13 Comparison of MPPT tech
hnique for ABC, gbest
and capacitance ‘C= 450uF’. The lowest THD is for line ABC, and PO
voltage AB (V_ABL) 11.88%, line voltagge BC (V_BCL)
11.19% and line voltage CA (V_CAL), 5.04% % Table 3: Output power of PV arra
ay at different irradiance:
Table 1: THD at different value of L and
dC
Power by diffeerent algorithms (W)
L C THD THD THD Irradiance
S.No. ABC Gbest
G ABC PO
(mH) (ȝF) (V_ABL) (V_BCL)) (V_CAL)
1 1.157 250 12.31% 11.66% 5.54% 950 196.002 196.012 146.201
2 1.157 450 12.26% 12.01% 5.48% 750 148.879 155.013 137.658
550 100.786 103.012 106.1
3 1.63 220 12.07% 11.40% 5.38%
250 28.346 29.048 30.003
4 1.63 300 12.01% 11.65% 5.38%
5 1.63 250 12% 11.46% 5.31%
VI. CONCLU
USION
6 1.63 280 11.87% 11.15% 5.19%
7 1.63 330 11.77% 11.01% 5.18% In this paper hybrid PV-BSU geeneration system has been
8 1.63 450 11.88% 11.19% 5.04% studied and useful results have beenb accumulated. MPPT
techniques based on GA has beeen compared and found
Table 2: Fundamental value of line voltagees at different competitive with conventional PO algorithm. GA based MPPT
value of L and C for PVA alone for PV generation works for local opptimal solution. DC/DC and
boost converter has been implemen nted successfully to obtain
L C ( V_ABL) (V_BCLL) (V_CAL) line voltage of 400V. Single section LC filter at output terminal
S.No.
(mH) (ȝF) rms rms rms eliminates harmonic components.
1 1.157 250 253.1 251.6 238.1 APPENDIIX
2 1.157 450 258.7 257.6 243.5
3 1.63 220 247.4 246.8 233 A. Photovoltaic Array specificatiions
4 1.63 300 250.5 249.9 236
Electron charge=1.602 × 10-19(C)
5 1.63 250 248.6 248 234.1 Boltzmann constant=1.38 × 10-23 J//oK.
6 1.63 280 249.8 248.9 235 Photocurrent=5A
7 1.63 330 251.8 250.9 236.9 Reverse saturation current of diode=
=.0002A
8 1.63 450 256.5 255.4 241.2

2015 International Conference on Power and Advanced Control Engineering (ICPACE) 5

Series resistance of cell=.001 Ÿ Incremental Conductance MPPT Methods for PV
Reference cell operating temperature=200C Systems," IEEE Journal of Photovoltaics, vol. 3, no. 3,
pp. 1070-1078, July 2013.
B. Battery specification [8] M.A.G. de Brito, L. Galotto, L.P. Sampaio, G. de
-1 Azevedo e Melo, and C.A. Canesin, "Evaluation of the
B= Exponential capacity (Ah) = 25
Main MPPT Techniques for Photovoltaic
R=Internal resistance (ohms) =.066667
Applications," IEEE Transactions on Industrial
E0 = Constant voltage (V) =421.5
Electronics, vol. 60, no. 3, pp. 1156-1167, March 2013.
Q = Battery capacity (Ah) =150
[9] M.A. Elgendy, B. Zahawi, and D.J. Atkinson,
A = Exponential voltage (V) =22 "Assessment of Perturb and Observe MPPT Algorithm
Implementation Techniques for PV Pumping
C. Boost converter2 specification Applications," IEEE Transactions on Sustainable
IGBT switch Energy, vol. 3, no. 1, pp. 21-33, Jan. 2012.
Capacitor=7500ȝF [10] D.L. King, W.E. Boyson, and J.A. Kratochavil,
Inductor =5ȝH “Photovoltaic array performance model”, Sandia
Resistance=.2ohm National Laboratories, Technical Report, SAND2004.
[11] M.M. El-Wakil, “Power plant technology”, McGraw-
D. MPPT Buck converter specification Hill, New York, 1984.
[13] T. Ishikawa, “Grid-connected photovoltaic power
Capacitor =4400ȝF systems: survey of inverter and related protection
Inductor =4ȝH equipments”, Report IEA (International Energy Agency)
MOSFET switch PVPS, 2002. http://www.iea-pvps.org
REFERENCES [12] Solarbuzz, “solar Cell Technologies”, 2010.
http://www.solarbuzz.com/technologies.htm
[1] S.K. Kollimalla, M.K. Mishra, "A Novel Adaptive P&O [14] R. Menzies, “Designing a solar power systems”, Proc.
MPPT Algorithm Considering Sudden Changes in the Solar Photovoltaic Energy Workshop'98, Monash
Irradiance", IEEE Transactions on Energy Conversion, University, Caulfield, 1998.
vol. 29, no. 3, pp. 602-610, Sept. 2014. [15] Bright Green Energy, “What type of solar PV panels are
[2] C. Cabal, L. Martínez-Salamero, L. Séguier, C. Alonso, available?”, 2009.
and F. Guinjoan, "Maximum power point tracking based http://www.wirefreedirect.com/Types_Solar_PV_Panels.
on slidingmode control for output-series connected asp
converters in photovoltaic systems," IET Power
Electronics, vol. 7, no. 4, pp. 914-923, April 2014. AUTHORS:
[3] M.A.A. Mohd Zainuri, M.A. Mohd Radzi, A.C. Soh, and Prakash Kumar: Presently he is PhD Scholar in Electrical
N.A. Rahim, "Development of adaptive perturb and Engineering Department, university college of Engg, Kota. He
observe-fuzzy control maximum power point tracking for has done M.Tech and B.tech in year 2012 and 2008
photovoltaic boost dc-dc converter," IET Renewable respectiverly. His research interest includes in Power System
Power Generation, vol. 8, no. 2, pp.183-194, March and Renewable energy.
2014.
[4] S.K. Kollimalla, M.K. Mishra, "Variable Perturbation Gaurav Jain: Presently he is PhD Scholar in Electrical
Size Adaptive P&O MPPT Algorithm for Sudden Engineering Department, university college of Engg, Kota. He
Changes in Irradiance," IEEE Transactions on has done M.Tech from Guru Nanak Dev Engineering College
Sustainable Energy, vol. 5, no. 3, pp. 718,728, July 2014. Ludhiana, India. His research interest includes in Power
[5] M.A. Elgendy, B. Zahawi, and D.J. Atkinson, System, Power Plant Engineering and Transmission pricing.
"Assessment of the Incremental Conductance Maximum Dheeraj Kumar Palwalia: Presently he is a Dy. Director
Power Point Tracking Algorithm," IEEE Transactions on (Acad) Rajasthan Technical University and Associate Prof. in
Sustainable Energy, vol. 4, no. 1, pp. 108-117, Jan. 2013. Electrical Engineering Department in University College of
[6] D.C. Jones, R.W. Erickson, "Probabilistic Analysis of a Engg., RTU, Kota, India. He received the Ph.D. degree from
Generalized Perturb and Observe Algorithm Featuring Indian Institute of Technology, Roorkee. His research interest
Robust Operation in the Presence of Power Curve includes in power quality, Power System and Transmission
Traps," IEEE Transactions on Power Electronics, vol. pricing.
28, no. 6, pp. 2912-2926, June 2013.
[7] D. Sera, L. Mathe, T. Kerekes, S.V. Spataru, and R.
Teodorescu, "On the Perturb-and-Observe and

6 2015 International Conference on Power and Advanced Control Engineering (ICPACE)