2015 Second International Conference on Mathematics and Computers in Sciences and in Industry

Identification of Solar Cell Parameters

with Firefly Algorithm

Mohamed LOUZAZNI1, Aurelian CRĂCIUNESCU2, El Hassan AROUDAM1, Alexandru DUMITRACHE3

1
Team Modeling and Simulation of Mechanical System, Faculty of Sciences, Abdelmalek Essaadi University,
Tetouan, Morocco.
2
Faculty of Electrical Engineering, University “Politehnica” of Bucharest,
Bucharest, Romania.
3
"Gheorghe Mihoc-Caius Iacob" Institute of Mathematical Statistics and Applied Mathematics of the Romanian Academy.
[email protected], [email protected], [email protected]

Abstract—Due to the non-linearity, multivariable and multi- for nonlinear and complex system, meta-heuristic optimization
modal features of current-voltage of solar cell models, the algorithms such as particle swarm optimization [9], genetic
conventional methods are incapable to estimating the parameters algorithm [10], simulated annealing [11], harmony search
of solar cell with high accuracy. Recently, the bio-inspired [12], pattern search [13], artificial bee swarm optimization
algorithms have attracted the attention. The firefly algorithm is algorithm [14], mutative-scale parallel chaos optimization
nature-inspired stochastic optimization algorithm is among the algorithm [15], chaotic asexual reproduction optimization [16]
most powerful algorithms in solving modern global optimization and bird mating optimizer [17] have been used to extract and
for nonlinear and complex system, based on the flashing patterns identify the parameter estimation issue. The meta-heuristic
and behavior of fireflies swarm. In this paper, a firefly algorithm algorithms have been suggested for the parameters extraction
is proposed to extract the parameters of single diode model solar
and becoming an important part of modern optimization.
cell from experimental I-V characteristics. The results obtained
by firefly algorithm are quite promising and outperform those The firefly algorithm as developed by Yang [18] is
found by the other methods. nature-inspired stochastic optimization algorithm is among the
most powerful algorithms in solving modern global
Keywords— Solar Cell, Conventional Methods, Bio-inspired optimization for nonlinear and complex system, based on the
Algorithms, Complex System, Firefly Algorithm flashing patterns and behavior of fireflies swarm. The firefly
algorithm has become an increasingly important tool of swarm
I. INTRODUCTION intelligence that has been applied in almost all areas of
The major challenge in solar cell is posed by its complex optimization, as well as engineering practice[19]. It uses a
and nonlinear current-voltage characteristics equation, which kind of randomization by searching for set solutions, inspired
are due to the varying operating temperature and radiation, by the flashing lights of fireflies in nature. In this paper, a
which contributes to the low energy conversion of the firefly algorithm based parameter estimation method for the
photovoltaic system. In research area, they are a several solar cell single diode is presented. Simulation and
mathematical models that describe the operation and electrical experimental results obtained by firefly algorithm are quite
behavior and many studies are focused on the modeling of promising and outperform and the feasibility of the proposed
solar cell and developing several electric models with different parameter estimation method.
level of complexity [1][2]. These models differ mainly by the
number of diodes, the shunt resistor (finite of infinite), the II. SOLAR CELL MODEL
ideality factor (fixed or variable) and the numerical methods The electrical behavior of solar cell is modeled by its
used to determine the unknown parameters. The exponential outputs current-voltage characteristic represent the ability and
nonlinearity of the current-voltage equation causes many keys to analysis the photovoltaic performance. A typical
difficulties in identification and extraction of the parameters model for a solar cell is shown in Fig.1, Therefore, the
[3]. Over the year, various parameters extraction techniques generated voltage can be represented by:
have been proposed to identify the optimal values of the
unknown parameters, generally the can be categorized into ⎡ + I P V .R s ⎤ V + I PV .R s
I pv = ⎡⎣ I sc + K i ( TC − T ref ) ⎤⎦
G V (1)
− I s ⎢exp( PV ) − 1⎥ − PV
analytical methods and numerical methods [4]. The analytical G r ef ⎣ a Vth ⎦ RP

methods have been used in [5][6][7][8] require continuity,

The first term is the short circuit current. The second term
convexity and differentiability conditions for being applicable,
contains the dark saturation current due to the recombination
and due of the nonlinearity of current-voltage makes the
analytical optimization techniques unable to effectively solve in the quasi-neutral region I s derived from the Shockley diode
the parameter identification problem. In recent years, because equation including a diode quality factor n to approximate the
of their great potentials in solving modern global optimization

978-1-4799-8673-6/15 $31.00 © 2015 IEEE 7

DOI 10.1109/MCSI.2015.37
effect of both diffusion and recombination in the space charge III. PARAMETER IDENTIFICATION AND OBJECTIVE
layer. FUNCTION

Where, ISC is the short circuit current at ambient temperature The unknown parameters are optimized and extracted with
Tref and reference irradiance intensity 1000 (W/m2) Gref, G meta-heuristic algorithm based to the optimization of the
is the irradiance intensity (W/m2), Ki is the cell’s short circuit objective function of the solar cell and minimizing. In the
current temperature coefficient at ambient temperature Tref. Is optimization algorithm, each solution is defined by a vector
the diode saturation current, Vth representing the thermal x = ⎡⎣ I ph I 0 a Rs Rsh ⎤⎦ during the optimization process; the
voltage and Rs the series resistance RP and the shunt objective function must be minimized with respect to the
resistance. limits of parameters x . The equation (6) is rewritten in the
following homogeneous equations.
f (V , I , x ) = 0 (7)
The objective function is the sum of the individual absolute
error (IAE), can be formulated by
20

∑ ( f (V , I
i =1
i i i , x )) = 0 (8)

Where Vi and Ii are ith measured current and voltage

respectively.
The objective function must minimize subject to:

Fig. 1. The single diode model of solar cell. g ( x1 ) = x 1 − 1

g (x2 ) = x2 − 1
g ( x3 ) = x3 − 2 (9)
The TC solar cell temperature is modeled as the flowing
g ( x 4 ) = x 4 − 0 .5
equation [7]:
g ( x 5 ) = x5 − 1 0 0
⎛ G ⎞
⎜ G ref ⎟⎟ (
T C = 3.12 + 0.25 ⎜ + 0 .8 89 T a ) − (1.3W S ) + 273 (2)
⎝ ⎠

Where Ta is the ambient temperature (°C) and Ws is the local Under the limit of the variable
wind speed (m/s).
0 ≤ x1 ≤ 1
The diode saturation current is formulated as follows: 0 ≤ x2 ≤ 1
3 ⎛ E g , ref E g ⎞ 1 ≤ x3 ≤ 2
⎛ T ⎞ ⎜
⎜ K T r ef
−
KT
⎟
⎟
(10)
I S = I RS ⎜ C ⎟ e ⎝ ⎠
0 ≤ x 4 ≤ 0 .5
⎜T ⎟ (3)
⎝ re f ⎠ 0 ≤ x5 ≤ 1 0 0
Where, Eg,ref is the band gap energy at Tref=298,16 K, IRS During the optimization, the objective function is minimized
characterizes diode reverse saturation current. The with respect to the parameter range, provided by the literature
semiconductor band gap energy Eg is are shown in Table 1.
T C2
E g (TC ) = 1 .1 7 − 4 .7 3 × 1 0 − 4 × (4)
TC + 6 3 6 TABLE I. UPPER AND LOWER RANGE OF THE SOLAR CELL PARAMETERS
Parameter Lower Upper
I ph 0 1

The current variation is less dependent on the shunt resistance, I0 0 1

and is more dependent on the series resistance.
a 1 2
⎡ + I PV .R s ⎤ V + I P V .R s
I p v = ⎡⎣ I sc + K i ( T C − T r ef ) ⎤⎦ (5)
G V
− I s ⎢ ex p ( P V ) − 1⎥ − PV 0 0.5
G ref ⎣ a V th ⎦ RP Rs

Rsh 0 100
⎡ V + I PV .R s ⎤ V + I PV .R s
I pv = I ph − I s ⎢ e x p ( PV ) − 1⎥ − PV (6)
⎣ a V th ⎦ RP

8
 IV. FIREFLY ALGORITHM Where xi , k is the kth component of the spatial coordinate xi
Firefly algorithm is a bio-inspired meta-heuristic algorithm for of ith firefly the movement of a firefly i is attracted to another
optimization problems. It was introduced in 2009 at more attractive firefly j and the update location is determined
Cambridge University by Yang [18]. The algorithm is inspired by
by the flashing behavior of fireflies at night. Firefly inspired
algorithms use the following three idealized rules: (i) All ⎛ 1⎞
fireflies are unisex which means that they are attracted to other xi +1 = xi + β 0 e
− λ rij2
(x j − xi ) + α ⎜ rand − ⎟
fireflies regardless of their sex; (ii) The degree of the ⎝ 2⎠ (17)
attractiveness of a firefly is proportion to its brightness,
thus for any two flashing fireflies, the less brighter one The first term is the current position of a firefly [20], the
will move towards the brighter one and the more second term is used for considering a firefly's attractiveness to
brightness means the less distance between two fireflies. If light intensity seen by adjacent fireflies and the third term is
there is no brighter one than a particular firefly, it will move used for the random movement of a firefly in case there are
randomly; (iii) the brightness of a firefly is determined by the not any brighter ones. The coefficient α is a randomization
value of the objective function [20]. parameter determined by the problem of interest, while rand is
a random-number generator uniformly distributed in the space
(0,1). In the implementation of the algorithm we will use
A. Attractiveness
β 0 = 0.1,α ∈[ 0 1] and the attractiveness or absorption
The intensity of light I is inversely proportional to the square coefficient λ = 1 which guarantees a quick convergence of
of the distance r [21]and expressed by the following the algorithm to the optimal solution.
equation:
The algorithm can be summarized as follows.
I ( r ) = I0 e − λ r
2
(12)
(1) Generate a random solution set x = ( x1 , x2 .....xk ) .
Where I 0 the intensity at the source point and the
(2) Compute intensity for each solution
attractiveness [22]is defined by member I = ( I1 , I2 .....I k ) .
β ( r ) = β 0 e− λr
2

(13) (3) Move each firefly i towards other brighter fireflies, and if
there is no other brighter firefly, move it randomly.
β is the light absorption coefficient and β 0 is the
attractiveness at r = 0 . (4) Update the solution set.
(5) Terminate if a termination criterion is fulfilled otherwise
However, computationally computing 1 / (1 + λ r 2 ) is easier go back to step 2.
−λr2
than e [18] and the intensity can be written as Schematically, the Firefly Algorithm can be summarized as
the pseudo code.
I0
I (r) =
1 + λr 2 (14)
Similarly, the attractiveness of a firefly can be defined as
follows:
β0
β (r) =
1 + λr 2 (15)

B. Distance and movement

We suppose a firefly is located at xi = x1i , x2i .....xki ( ) is
brighter than another firefly located at x j = x , x .....x ( 1
j
2
j
k
j
) , the
firefly located at xi will move towards x j . The distance
between any two fireflies i and j at xi and x j is the Cartesian
distance given by [18]as follows:
d

∑( x − x j ,k )
2
rij = xi − x j = i ,k (16)
k

Fig. 2. Pseudo code of the Firefly Algorithm

9
 V. SIMULATION RESULTS relative error of firefly algorithm is small than HM, BMO,
Firefly algorithm based parameter estimation method is used CARO and MPCOA.
to extract the five parameters of solar cell. The solar radiation
and module temperature are fixed and the I-V curve is TABLE III. DIFFERENT CURRENT CALCULATED USING DIFFERENT
measured. The data is obtained from [5], where a 57 mm ALGORITHMS
diameter commercial (R.T.C France) silicon solar cell (at 33 Item FA HS[12] BMO[24] CARO[16] MPCOA[15]
°C). The algorithms are coded and executed in the Matlab 1 0.7606 0.760293 0.760153 0.760121 0.760117
environment to identify the cell parameters using the single 2 0.7593 0.759181 0.759058 0.759031 0.759037
diode model. The parameters are extracted with other meta- 3 0.7584 0.758155 0.758048 0.758026 0.758041
4 0.7573 0.757193 0.757099 0.757082 0.757105
heuristic algorithms, for studying the correctness and 5 0.7565 0.756232 0.756152 0.756139 0.756169
effectiveness of the algorithm utilized the results are 6 0.7554 0.755167 0.755099 0.755091 0.755125
compared. 7 0.7538 0.753734 0.753676 0.753674 0.753712
8 0.7517 0.751446 0.751399 0.751408 0.751441
Extracted optimal parameters values for the single diode 9 0.7476 0.747394 0.747357 0.747378 0.747402
10 0.7405 0.740129 0.740102 0.740149 0.740145
model by the firefly algorithm are compared with those found 11 0.7276 0.727416 0.727395 0.727421 0.727434
by several other techniques based on the same data, they are; 12 0.7074 0.706963 0.706945 0.707016 0.706975
harmony search-based algorithms (HS) [23], bird mating 13 0.6775 0.675302 0.675281 0.675325 0.675300
14 0.6313 0.630898 0.630868 0.630799 0.630875
optimizer (BMO) [24], chaotic asexual reproduction 15 0.5724 0.572110 0.572067 0.571959 0.572065
optimization (CARO) [16] and mutative-scale parallel chaos 16 0.4998 0.499537 0.499482 0.499622 0.499474
17 0.4139 0.413554 0.413490 0.413646 0.413481
optimization algorithm (MPCAO) [15]are summarized in 18 0.3175 0.317286 0.317223 0.317192 0.317214
Table II. 19 0.2123 0.212161 0.212111 0.212126 0.212105
20 0.1030 0.102756 0.102731 0.102621 0.102728

TABLE II. COMPARISON AMONG DIFFERENT PARAMETER EXTRACTION

Item FA HS [12] BMO CARO MPCOA TABLE IV. COMPARISON OF RELATIVE ERROR OF EACH MEASUREMENT
[23] [16] [15] (RE)

Rs ( Ω ) 0.04801 0.03663 0.03636 0.03644 0.03635 Item

MPCO[15]
FA HS[12] BMO[24] CARO[16]
Rsh ( Ω ) 53.2704 53.5946 53.8716 53.0893 54.6328
1 -0.000131 0.000272 0.000456 -0.00049 0.00050
I ph ( A ) 0.7124 0.76070 0.76077 0.76079 0.76073 2 0.000921 0.001078 0.001239 0.00127 0.00126
3 0.000790 0.001113 0.001254 0.00128 0.00126
I 0 ( A) 0.9766 0.30495 0.32479 0.31724 0.32655 4 -0.000396 -0.000250 -0.00013 -0.00011 -0.00014
5 0.000660 0.001015 0.001121 0.00114 0.00109
a 1.4622 1.47538 1.48173 1.48168 1.48168 6 0.000132 0.000441 0.000531 0.00054 0.00049
7 0.000265 0.000353 0.000429 0.00043 0.00038
8 -0.001598 -0.00126 -0.00120 -0.00121 -0.00125
9 -0.001473 -0.00120 -0.00115 -0.00118 -0.00121
In order to verify the quality of the results, we calculus the 10 -0.002708 -0.00221 -0.00217 -0.00223 -0.00222
11 0.000549 0.000802 0.000831 0.00079 0.00077
relative error and the individual absolute error for each 12 -0.001273 -0.00066 -0.00063 -0.00073 -0.00067
measurement by 13 -0.002960 0.000293 0.000324 0.00026 0.00030
14 0.001107 0.001744 0.001791 0.00190 0.00178
⎛I −I ⎞ 15 0.001047 0.001553 0.001628 0.00182 0.00163
RE = ⎜ i cal ⎟ (18) 16
17
-0.001603
-0.002179
-0.001080
-0.001340
-0.00097
-0.00119
-0.00125
-0.00157
-0.00095
-0.00116
⎝ Ii ⎠ 18 -0.003159 -0.002480 -0.00228 -0.00219 -0.00226
19 -0.001415 -0.000760 -0.00053 -0.00059 -0.00049
IAE = I i − I cal (19)
20 0.004830 0.007188 0.007433 0.00849 0.00745

Where I cal is the current calculated for each Vi by solving the -3 Relative error of each measurement by FA, HS, BMO, CARO and MPCO
x 10
equation (6) . 10
FA
8
HS
The Table III present the current calculated for firefly
6 BMO
algorithm compared with harmony search-based algorithms, CARO
Relative error

bird mating optimizer, chaotic asexual reproduction 4 MPCO

optimization and mutative-scale parallel chaos optimization 2
algorithm. The statistical analysis results using the
0
fundamental measures of accuracy such as relative error and
individual absolute error are given in table IV and V -2

respectively. -4
0 2 4 6 8 10 12 14 16 18 20
Measurement
The figure 3 shows the compared relative error curve of firefly
algorithm with other meta-heuristic algorithms; we can see the
Fig. 3. RE of each measurment

10
 TABLE V. COMPARISON OF INDIVIDUAL ABSOLUTE ERROR OF EACH 0.35
MEASUREMENT ( IAE) FA
0.3 Real data
Item
MPCO[15]
FA HS[12] BMO[24] CARO[16] 0.25

Power (W)
1 0.0000996255 0.0002 0.0012 0.000379 0.000383 0.2
2 0.0006999600 0.0008 0.0134 0.000969 0.000963
3 0.0005996100 0.0008 0.0140 0.000974 0.000959 0.15
4 0.0002997720 0.0002 0.0003 0.000082 0.000105
5 0.0004996200 0.0008 0.0114 0.000861 0.000831
6 0.0000997260 0.0003 0.0023 0.000409 0.000375 0.1
7 0.0001998100 0.0003 0.0015 0.000326 0.000288
8 0.0011992990 0.0009 0.0151 0.000908 0.000941 0.05
9 0.00109959450 0.0009 0.0138 0.000878 0.000902
10 0.00199985800 0.0016 0.0480 0.001649 0.001645 0
11 0.00039967200 0.0006 0.0067 0.000579 0.000566 0 0.1 0.2 0.3 0.4 0.5 0.6
12 0.00089937450 0.0005 0.0042 0.000516 0.000475 Voltage (V)
13 0.00199948000 0.0002 0.0009 0.000175 0.000200
14 0.00069962400 0.0011 0.0312 0.001201 0.001125 Fig. 6. Comparison results from the real data and the solar cell module
15 0.00059993100 0.0009 0.0253 0.001041 0.000935
16 0.00079989700 0.0005 0.0105 0.000622 0.000474
17 0.00089992700 0.0006 0.0159 0.000646 0.000481
18 0.00099982350 0.0008 0.0565 0.000692 0.000714 VI. CONCLUSION
19 0.00029998000 0.0002 0.0040 0.000126 0.000105
20 0.00049990500 0.0007 0.5343 0.000879 0.000772 The aims of this paper deals about the application of
the firefly algorithm to extract the five parameters of the solar
cell single diode model. The five parameters namely the photo
Figure 4 shows each measurement respectively using FA and current, diode saturation currents, diode ideality factors, series
comparing with HM, BMO, CARO and MPCOA. resistance and shunt resistance are extracted using firefly
3
x 10
-3
algorithm. The firefly algorithm offers several advantages
FA such as accuracy of solution, simple to be used with objective
2.5 HS function and speed of convergence. The comparison results
BMO
CARO
with other meta-heuristic algorithm shows the accuracy and
2
MPCO the advantage of the algorithm.
IAE

1.5

1
ACKNOWLEDGMENTS
This work is in part supported by the “Agence
0.5
Universitaire de la Francophonie (AUF)”. The first author
0 would like to acknowledge a support from ministry of foreign
0 5 10 15 20
Measurement
affairs of Romania for the “Eugen Ionesco” Scholarship. Also,
partially, this work was realized through the Partnership
Fig. 4. IAE of each measurment program in priority domains - PN II, developed with support
from ANCS CNDI - UEFISCDI, project no. PN-II-PT-PCCA-
The figure 5 and 6 shows the I-V and P-V experimental 2011-3.2-1670.
characteristics plots and the fitted curve with the parameters
obtained by firefly algorithm. The calculated values are REFERENCE
accordance with the measured data observably and they are [1]G. Kumar et A. K. Panchal, « A non-iterative technique for determination
good agreement for most of the extracted parameters. of solar cell parameters from the light generated I-V characteristic », Journal
of Applied Physics, vol. 114, no 8, p. 084904, 2013.
0.8
[2]H. Fathabadi, « Novel neural-analytical method for determining
silicon/plastic solar cells and modules characteristics », Energy Conversion
0.7 and Management, vol. 76, p. 253̻259, déc. 2013.
[3]Y. Chen, X. Wang, D. Li, R. Hong, et H. Shen, « Parameters extraction
0.6 from commercial solar cells I–V characteristics and shunt analysis », Applied
Real data
FA
Energy, vol. 88, no 6, p. 2239̻2244, juin 2011.
Current (A)

0.5 [4]J. Appelbaum et A. Peled, « Parameters extraction of solar cells – A

comparative examination of three methods », Solar Energy Materials and
0.4 Solar Cells, vol. 122, p. 164̻173, mars 2014.
[5]T. EASWARAKHANTHAN, J. BOTTIN, I. BOUHOUCH, et C.
0.3 BOUTRIT, « Nonlinear Minimization Algorithm for Determining the Solar
Cell Parameters with Microcomputers », International Journal of Solar
0.2
Energy, vol. 4, no 1, p. 1̻12, janv. 1986.
0.1
[6]D. S. H. Chan, J. R. Phillips, et J. C. H. Phang, « A comparative study of
0 0.1 0.2 0.3 0.4 0.5 0.6 extraction methods for solar cell model parameters », Solid-State Electronics,
Voltage (V) vol. 29, no 3, p. 329̻337, mars 1986.

Fig. 5. Comparison results from the real data and the solar cell module

11
[7]A. Jain et A. Kapoor, « Exact analytical solutions of the parameters of real [17]A. Askarzadeh et L. dos Santos Coelho, « Determination of photovoltaic
solar cells using Lambert W-function », Solar Energy Materials and Solar modules parameters at different operating conditions using a novel bird
Cells, vol. 81, no 2, p. 269̻277, févr. 2004. mating optimizer approach », Energy Conversion and Management, vol. 89,
[8]H. Saleem et S. Karmalkar, « An Analytical Method to Extract the Physical p. 608̻614, janv. 2015.
Parameters of a Solar Cell From Four Points on the Illuminated Curve », IEEE [18]X.-S. Yang, Nature-inspired Metaheuristic Algorithms. Luniver Press,
Electron Device Letters, vol. 30, no 4, p. 349̻352, avr. 2009. 2010.
[9]M. Ye, X. Wang, et Y. Xu, « Parameter extraction of solar cells using [19]I. Fister, I. Fister, X.-S. Yang, et J. Brest, « A comprehensive review of
particle swarm optimization », Journal of Applied Physics, vol. 105, no 9, p. firefly algorithms », Swarm and Evolutionary Computation, vol. 13, p. 34̻46,
094502, 2009. déc. 2013.
[10]M. Zagrouba, A. Sellami, M. Bouaïcha, et M. Ksouri, « Identification of [20]X.-S. Yang, Engineering optimization: an introduction with metaheuristic
PV solar cells and modules parameters using the genetic algorithms: applications. Hoboken, N.J: John Wiley, 2010.
Application to maximum power extraction », Solar Energy, vol. 84, no 5, p. [21]I. Fister, X.-S. Yang, J. Brest, et I. Fister, « Modified firefly algorithm
860̻866, mai 2010. using quaternion representation », Expert Systems with Applications, vol. 40,
[11]K. M. El-Naggar, M. R. AlRashidi, M. F. AlHajri, et A. K. Al-Othman, no 18, p. 7220̻7230, déc. 2013.
« Simulated Annealing algorithm for photovoltaic parameters identification », [22]A. Kazem, E. Sharifi, F. K. Hussain, M. Saberi, et O. K. Hussain,
Solar Energy, vol. 86, no 1, p. 266̻274, janv. 2012. « Support vector regression with chaos-based firefly algorithm for stock
[12]A. Askarzadeh et A. Rezazadeh, « Parameter identification for solar cell market price forecasting », Applied Soft Computing, vol. 13, no 2, p. 947̻958,
models using harmony search-based algorithms », Solar Energy, vol. 86, no févr. 2013.
11, p. 3241̻3249, nov. 2012. [23]M. R. AlRashidi, M. F. AlHajri, K. M. El-Naggar, et A. K. Al-Othman,
[13]M. F. AlHajri, K. M. El-Naggar, M. R. AlRashidi, et A. K. Al-Othman, « A new estimation approach for determining the I–V characteristics of solar
« Optimal extraction of solar cell parameters using pattern search », cells », Solar Energy, vol. 85, no 7, p. 1543̻1550, juill. 2011.
Renewable Energy, vol. 44, p. 238̻245, août 2012. [24]M. F. AlHajri, K. M. El-Naggar, M. R. AlRashidi, et A. K. Al-Othman,
[14]A. Askarzadeh et A. Rezazadeh, « Artificial bee swarm optimization « Optimal extraction of solar cell parameters using pattern search »,
algorithm for parameters identification of solar cell models », Applied Energy, Renewable Energy, vol. 44, p. 238̻245, août 2012.
vol. 102, p. 943̻949, févr. 2013. [25]A. Askarzadeh et A. Rezazadeh, « A new heuristic optimization algorithm
[15]X. Yuan, Y. Xiang, et Y. He, « Parameter extraction of solar cell models for modeling of proton exchange membrane fuel cell: bird mating optimizer »,
using mutative-scale parallel chaos optimization algorithm », Solar Energy, Int. J. Energy Res., vol. 37, no 10, p. 1196̻1204, août 2013.
vol. 108, p. 238̻251, oct. 2014. [26]A. Askarzadeh et A. Rezazadeh, « Parameter identification for solar cell
[16]X. Yuan, Y. He, et L. Liu, « Parameter extraction of solar cell models models using harmony search-based algorithms », Solar Energy, vol. 86, no
using chaotic asexual reproduction optimization », Neural Comput & Applic, 11, p. 3241̻3249, nov. 2012.
p. 1̻13, déc. 2014. [27]A. Askarzadeh et A. Rezazadeh, « Extraction of maximum power point in
solar cells using bird mating optimizer-based parameters identification
approach », Solar Energy, vol. 90, p. 123̻133, avr. 2013.

12