Renewable Energy 216 (2023) 119102

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

An efficient method for predicting PV modules performance based on the

two-diode model and adaptable to the single-diode model
Kawtar Tifidat *, Noureddine Maouhoub
Department of Physics, Laboratory of Materials, Signals, Systems, and Physical Modeling, Faculty of Sciences, Ibn Zohr University, B.P. 8106, 80000, Agadir, Morocco

A R T I C L E I N F O A B S T R A C T

Keywords: This paper introduces a simplified and accurate modeling approach to model photovoltaic modules. The current
PV module method uses a hybrid technique combining numerical and analytical approaches to predict the current-voltage
I–V characteristic characteristics of PV generators by calculating the parameters of the double-diode model. First, to reduce the
Parameter extraction
complexity of the current equation and minimize the calculation time, some simplifications reducing the research
Seven parameter
Double-diode model
space from seven to only five unknowns are taken into consideration. Second, three of these five parameters are
Energy conversion extracted using simple analytical equations derived from the output current’s equation based on the available
values of the typical points on the manufacturer’s datasheet. Then, the remaining two parameters are extracted
using a combination of numerical approach and slope adjustment technique at the short-circuit point. The
technique is adapted to the single-diode model in order to study the effect of the second diode on the simulating
accuracy. The effectiveness is tested over six PV modules matching different PV technologies, and the results are
compared with a great number of modeling methods based on different approaches, and the adjustment tech­
nique has shown the highest accuracy levels (RMSE values less than 0.045A) at the least computational times
(compilation time less than 0.084 s).

energy has attracted many researchers throughout the world to ensure

the best benefit of sunlight. The interest was devoted to several axes
1. Introduction concerning the development of new photovoltaic modules and achieving
the most optimal way of their employment [5,6]. To that end, modeling
To meet the growing energy requirements implied by today’s tech­ photovoltaic generators and simulating their performance has also
nological development, nowadays, from an energy point of view, fossil emerged as an important task to be done to facilitate and offer more
fuels remain the most effective form of energy. And that makes them the opportunities to manipulate their device with lower costs [7,8].
most exploited source [1]. However, the massive widespread use of this Considering that achieving efficient and fast PV modeling allows the
last kind of energy has a considerable number of adverse effects. Burning prediction of generated electricity, the estimation of the maximum
fossil fuels causes the greenhouse effect by dropping large quantities of output power under outdoor conditions, and makes the comparison of
glasshouse gas into the atmosphere. And that directly implies the earth’s several PV array setups less expensive, modeling photovoltaic genera­
temperature rising and the pollution of the air leading to several health tors is considered a required task for performance assessment and the
diseases [2]. Moreover, the world’s reserves of fossil fuels are limited maximum power point tracking [9,10].
and concentrated only in some countries. Thus, their overexploitation Photovoltaic modeling is founded on equivalent electrical circuits.
has caused a rapid decrease in their quantities leading to economic The single-diode circuit model (SDM) and the double-diode circuit
problems, especially for non-energy-producing countries [3]. In the last model (DDM), are the most adopted electrical models in the literature.
few decades, energies such as; solar energy, wind energy, and biomass The main parameters of the single-diode model are the light-generated
energy, have emerged and are considered the alternatives that combine current IL, the shunt resistance Rsh, the series resistance Rs, the reverse
availability, cleanness, safety, and renewability [4]. saturation current Is, and the ideality factor a [11]. For the double-diode
The renewable energy revolution of today highlights photovoltaic model, in addition to the five parameters of the SDM another reverse
energy as an important resource and the most appealing one to meet saturation current Is2, and ideality factor a2 are added [12]. The wide
today’s energy requirements. The importance given to this kind of

* Corresponding author.
E-mail address: [email protected] (K. Tifidat).

https://doi.org/10.1016/j.renene.2023.119102
Received 20 October 2022; Received in revised form 25 July 2023; Accepted 30 July 2023
Available online 3 August 2023
0960-1481/© 2023 Elsevier Ltd. All rights reserved.
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

Nomenclature IGSK Improved Gaining-Sharing Knowledge Algorithm

DDM Double-Diode Model
a Ideality factor for the SDM KB(J/K) Boltzmann constant equals 1.38 × 10− 23 J/K
a1 Ideality factor of D1 for the DDM Ns Number of PV module’s cells in series
a2 Ideality factor of D2 for the DDM P (W) Output power
AM Air mass constant equals 1.5 PC,max(W) Calculated maximum power
G (W/m2) Received irradiance PM,max(W) Measured maximum power
I (A) Terminal current q (C) Electron charge equals 1.6021 × 10− 19C
Ical (A) Calculated current T(◦ C) PV generator’s temperature
IL(A) Photo-generated current VT (V) Thermal voltage
Is (A) Reverse saturation current for the SDM Vmp (V) Maximum power point’s voltage
IMeas(A) Measured current Voc (V) Open-circuit voltage
Imp(A) Maximum power point’s current V (V) Terminal voltage
Is1 (A) Reverse saturation current of D1 for the DDM KIsc (A/◦ C) Temperature coefficient of Isc
Is2 (A) Reverse saturation current of D2 for the DDM MLBSA Multiple Learning Backtracking Search Algorithm
MPP Maximum Power Point
Abbreviation RMSE Root Mean Square Error
AE Absolute Error SDM Single-Diode Model
AM Air Mass STC Standard Test Conditions
IJAYA Improved JAYA Algorithm

reliance on these two last models refers to their simplicity and modeling only for the ideal solar cells, and in this case, the resistive losses are
accuracy [13]. On the first hand, by neglecting the losses caused by the supposed to be totally absent. Thus, these assumptions are invalid when
recombination phenomenon occurring in the depletion region, the modeling real PV generators. Moreover, in Ref. [26], it has been accepted
one-diode equivalent circuit model avoids the calculation’s complexity that the two reverse saturation currents of the DDM are connected by an
and keeps offering sufficient efficiency, as well, thus it is widely selected approximate predetermined relation. Notwithstanding that the latter
by the authors [14,15]. On the other hand, the two-diode equivalent approximation does not have any physical significance or proof, it was
circuit model that takes into account the recombination losses allows first used by the authors in Ref. [27] and adopted later by several authors
more modeling precision but raises the prediction challenge by to simplify the needed calculations for the identification [28]. In addition
increasing the number of unknown parameters to calculate from five to to the approximation used for the two reverse saturation currents, in
seven [16,17]. Ref. [28], the values of the two ideality factors were taken directly for
Based on several approaches, various methods differing in their de­ each PV module from the two separate works in Refs. [29,30] which are
gree of accuracy and complexity were proposed in the literature to based on different extraction approaches. The light-generated current’s
predict PV generators’ parameters and estimate their performance. The value was also extracted by Yahya-Khotbehsara et al., using an additional
approaches introduced in the literature can be divided into two main assumption admitting the IL=Isc equality [28]. Then, after neglecting the
categories, analytical and numerical techniques. Given the non-linearity series resistance versus the shunt resistance in some calculation stages for
nature of the current equation, the analytical methods are based on some simplifications, it was extracted by the authors in Ref. [28] using an
assumptions to facilitate the calculations and guarantee sufficient pre­ iterative process based on an equation adjustment technique proposed
cision of the estimation [18,19]. The numerical methods include the use first by Majdoul et al., for the SDM [31].
of numerical resolution algorithms to solve a system of non-linear In the attempt to identify the parameter’s values of the DDM with a
equations derived from the current equation, iterative approaches, and lower degree of complexity, Ishaque et al., [32] assume the Is1 = Is2
optimization techniques based on meta-heuristic algorithms and data equality, which was adopted later in several works [33,34]. However,
fitting [20,21]. The major inconvenience of the numerical resolution is the reverse saturation current is highly dependent on temperature level
to find the initial values, especially for the saturation currents, the shunt since it is a leakage current, while the second saturation current is
resistance, and the ideality factors. The iterative methods find their derived from the second diode; thus, it models the electron-hole
limitation when the stopping constraints and the increment steps are not recombination current. Therefore, Is2 takes higher values than Is1, and
well chosen or when the number of loops increases. The optimization assuming the equality of both currents has no physical sense [35]. In
techniques generally require the availability of the experimental I–V Refs. [32,33], the light-generated current is assumed to be equal to the
characteristics, which limits the tasks devoted to the PV modeling, in short circuit current, and the first ideality factor is taken equal to 1 and
addition to the requirement of selecting accurate values for the opti­ the second ideality factor takes values greater than 1.2 directly assigned
mization parameters in the case of heuristic algorithms [22–25]. as fixed values to these two last parameters. For the works in Ref. [34],
Kumar et al., have based on a purely numerical approach to propose the two ideality factors are supposed to be equal and take values near 1.
a method applied for the SDM and DDM. The principle of this last is the However, this approximation has no physical significance because the
extraction of five equations for the SDM and seven equations for the first ideality factor measures the resemblance between the diode’s
DDM based on many approximations to be solved using the “fsolve” characteristics and those of an ideal diode, and it is generally close to 1
function of MATLAB software. Then, derive five and seven equations while, the second ideality factor corresponds to the recombination
using more approximations to get the required initial values for the phenomenon and takes 2 as a value when the phenomenon dominates
numerical resolution [17]. However, the complete reliance on the nu­ [35]. The series resistance and the shunt resistance are extracted in Refs.
merical resolution of equations’ systems, which all require initial [32,33] using an iterative loop based on the maximum power point
guessed values, highly affects the convergence of the resolution and adjustment, the technique which was first used by Villalva et al., [36] for
demands tedious calculations and more assumptions for simplification. the SDM.
The authors in Ref. [26] have assumed that the series resistance equals Zhuo et al., [37] have adopted the assumption admitting the equality
zero and the shunt resistance is infinite. However, these lasts are valid of both saturation currents. Then, based on Shockley diffusion theory, 1

2
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

was assigned to the diode-ideal diode similarity factor and the value 2 to SDM and Is1 and Is2 for the DDM) and the light-generated currents of
the recombination factor [38]. Next, the series resistance extraction both electrical models. The second method is the numerical approach
went through tedious calculations and was done in the final step using based on a fast iterative process to calculate the Rsh and Rs values for
an iterative process based on the mountain climbing algorithm. By both DDM and SDM. The extraction of the shunt resistance is done by
contrast, recently, to avoid the fixed values used for the two ideality using a new slope adjustment technique at the linear zone of the I–V
factors, Orioli et al., [39] and Chennoufi et al., [35] have chosen to in­ curve. Moreover, to guarantee high rapidity of convergence of the slope
crease the number of the used loops for the identification. Chennoufi searching towards the accurate solution, the search interval is defined
et al., [35] have proposed two methods; in the first case, the two fixed with great precision around the approximate value of Rsh. The increment
values a1 = 1 and a2 = 2 have been used. Then, the Rs value was steps are also defined and generalized using a new method devoted to
extracted using the Brent method. In the second case, to improve the selecting the increment steps relative to the initially found value of Rsh,
results, two successive iterative processes were used to extract the two by taking into consideration the small error that can be committed when
ideality factors’ values and the series resistance based on error mini­ extracting the approximate value of Rsh. The starting value is deter­
mization at the MPP. As a result, the iterative process provides almost 1 mined based on the I–V curve’s slope calculation at the linear zone.
for the first ideality factor and 2 for the second [6,35]. In this case, the Then, by utilizing the found value of Rsh, Rs is calculated using the Brent
use of two separate loops highly affects the required execution time for hybrid root-finding method [45].
the estimation [6]. Orioli et al., [39] have proposed an iterative tech­ The method in the current work is applied to different PV generators
nique based on three sequent loops to get the values of Rs, a1, and a2, of various technologies, including mono-crystalline technology, poly-
successively. Indeed, increasing the number of iterative processes may crystalline technology, and thin film technology. The effectiveness test
lead to more accurate results but amplifies the convergence time, which for the proposed method is done based on the calculation of some sta­
can be considered a negative point for real-time applications [6]. tistical indicators such as the RMSE (Root Mean Square Error), as well as
To extract the values of the electrical parameters, latterly, several the convergence speed. Then, the obtained values of these last indicators
optimization techniques employing the heuristic algorithms were also are evaluated against some of the well-known and most reviewed
introduced in the literature, for example, the use of the JAYA and methods in this field.
improved JAYA (IJAYA) algorithms [40], the multiple learning back­ In an attempt to organize the present paper, the SDM and the DDM
tracking search algorithm (MLBSA) [41], and the improved are detailed in section 2, the proposed method’s theory is introduced in
gaining-sharing knowledge algorithm (IGSK) [42]. However, these section 3, section 4 is devoted to the results and the test of the effec­
kinds of optimization techniques are time-consuming due to their tiveness, and section 5 gives the conclusion of the works.
counting on huge stochastic populations. Moreover, they require having
preliminary knowledge about the variation interval of each parameter to 2. Equivalent circuit of PV modules
extract, the thing that cannot always be available. Furthermore, when­
ever these algorithms are not well initialized, or if the intervals of Photovoltaic modules are constructed from the combination of
variation were not properly chosen, they become unable to converge identical solar cells arranged in series and parallel to hit the required
and produce satisfactory results. In addition to the selection of initial currents and voltages. To model a single solar cell, two models are
values for the model’s parameters, these last algorithms require also a selected in this work. The first model shown in Fig. 1 is the one-diode
good predefinition of their own linked parameters, which varies from model in which the used diode goes back to the juxtaposition of two
one algorithm to another [43]. semiconductors of different doping nature inside the solar cell, and it
The objective of this paper is to propose a hybrid prediction tech­ gives rise to two electrical parameters; the ideality factor a and the
nique that is adaptable to both DDM and SDM models for estimating the reverse saturation current Is. The model consist also of the generator of
performance of PV generators. The ideality factors’ values are first current that models the light-generated current to which the current IL
determined using the Shockley diffusion theory [44]. According to this parameter is associated. The two last electrical parameters are Rs which
theory, the diode-ideal diode similarity factor is equal to 1 and the is the series resistance that models the losses due to the joule effect, and
recombination ideality factor’s value is 2. Then, the proposed method the shunt resistance Rsh modeling the leaks through the PV cell’s PN-
uses only the values of the I–V curve’s remarkable points to compute the junction related to the imperfection of the used materials [20,23].
other parameters. In fact, the identification methodology is divided into The non-linear mathematical equation linking the terminal current I
two extraction approaches using the I–V equation at the short circuit of a PV generator to the terminal voltage V is extracted based on the
current, the maximum power point (MPP), and the open-circuit voltage, single-diode model using the mesh law applied to the circuit of Fig. 1,
in addition to the use of the current’s derivative with respect to the and it is written as follows [46]:
voltage (V) at the MPP. The first approach is an analytical method based [ ( ) ]
IRs + V
on matrix resolution, which is used to calculate the parameters that are I = IL − Is exp − 1 − Gsh (IRs + V) (1)
aVT Ns
not obvious to be initialized, including the saturation currents (Is for the

Fig. 1. Single-diode model.

3
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

Fig. 2. Double-diode model.

where Ns is the number of series cells of the PV generator, the thermal For (I=Imp and V=Vmp)
[ ( ) ] [ ( ) ]
voltage VT is given as VT=(KB/q) × T, where KB is the Boltzmann con­ Imp Rs + Vmp Imp Rs + Vmp
stant, q is the electron charge, and T is the PV cell’s back surface tem­ Imp = IL − Is1 exp
a1 VT Ns
− 1 − Is2 exp
a2 VT Ns
− 1
perature. Gsh is the shunt admittance defined as Gsh = 1/Rsh. ( )
− Gsh Imp Rs + Vmp (4)
The second selected electrical model is the double-diode model
shown in Fig. 2. In addition to the effects considered in the SDM, the For (I = 0 and V=Voc)
DDM takes into consideration the recombination phenomenon occurring [ ( ) ] [ ( ) ]
Voc Voc
in the PV cell’s space charge region. This last effect is modeled by a IL − Is1 exp − 1 − Is2 exp − 1 − Voc Gsh = 0 (5)
a1 VT Ns a2 VT Ns
second diode added to the SDM to form the DDM with two more pa­
rameters. Hence, the two-diode model contains seven unknown pa­ By adopting the notations of Eq. (6) below:
rameters to be determined (IL, Is1, Is2, a1, a2, Rs, and Rsh), where a1 and ( ) ( )
Isc Rs Imp Rs + Vmp
Is1 are, respectively, the ideality factor and the reverse saturation cur­ Csc1,2 = exp − 1 ; Cmp1,2 = exp − 1;
rent corresponding to the first diode, and a2 and Is2 the ideality factor a1,2 VT Ns a1,2 VT Ns
( ) (6)
and the reverse saturation current corresponding to second diode [27, Coc1,2 = exp
Voc
− 1
28]. a1,2 VT Ns
Based on the double-diode model, the current equation of a PV Eqn 3, 4 and 5 can be managed and combined into a system of
generator is extracted using the mesh law applied to the circuit in Fig. 2, equations, which is equivalent to the equation (7).
and it is written as follows: ⎧
[ ( ) ] [ ( ) ] ⎪ Isc Rs
IRs + V IRs + V ⎪
⎪ IL − Csc,1 Is1 − Csc,2 Is2 = Isc +
⎪
⎪ Rsh
I = IL − Is1 exp − 1 − Is2 exp − 1 − Gsh (IRs + V) ⎪
⎪
a1 VT Ns a2 VT Ns ⎪
⎨ Voc
(2) IL − Coc,1 Is1 − Coc,2 Is2 = (7)
⎪
⎪ Rsh
⎪
⎪ ( )
⎪
⎪ Imp Rs + Vmp
3. Parameter extraction strategies ⎪
⎪
⎩ IL − Cmp,1 Is1 − Cmp,2 Is2 = Imp +
Rsh
To simulate PV generators’ performance, this section introduces an The system of Eq. (7) can be rearranged and written as follows:
extraction technique to first calculate the unknown parameters, which
are the important keys to PV simulation. The proposed technique is ADDM × λDDM = βDDM (8)
characterized by its flexibility, so that it can be adopted for both PV
where:
models; the DDM and the SDM. The issue that parameter extraction
⎛ ⎞
often confronts is the problem of negative values which have no physical 1 − Csc,1 − Csc,2
significance. To avoid this last issue, the calculation methodology is ADDM = ⎝ 1 − Coc,1 − Coc,2 ⎠ (9)
divided into two parts. The first part is the analytical calculation of the 1 − Cmp,1 − Cmp,2
light-generated current and the reverse saturation currents. The second ⎛ ⎞
part contains the adjustment of the slope in the vicinity of the short- Isc +
Isc Rs
circuit current to extract the shunt resistance and the numerical ⎜
⎜ Rsh ⎟
⎟
⎜ ⎟
extraction of the series resistance. ⎜ Voc ⎟
βDDM = ⎜
⎜
⎟
⎟ (10)
⎜ Rsh ⎟
⎜ ( )⎟
3.1. Analytical methodology ⎝
Imp +
Imp Rs + Vmp ⎠
Rsh
3.1.1. Double-diode model ⎛ ⎞
For the DDM, to get the values of IL, Is1, and Is2, analytically, the IL (Rs )
current equation gotten from the DDM is evaluated at the mean points of
⎝
λDDM = Is1 (Rs ) ⎠ (11)
Is2 (Rs )
the I–V characteristic and written as follows:
For (I=Isc and V = 0) The Eq. (12) below is solved to get the values of the three parameters
[ (
Isc Rs
) ] [ (
Isc Rs
) ] of the DDM; IL, Is1, and Is2.
Isc = IL − Is1 exp − 1 − Is2 exp − 1 − Gsh Isc Rs (3)
a1 VT Ns a2 VT Ns 1
λDDM = A−DDM × βDDM (12)

4
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

( ) ( ) ( )
3.1.2. Single-diode model dI Is1 dI IRs + V Is2 dI
=− Rs + 1 × exp − Rs + 1
For the SDM the same analytical procedure introduced for the DDM dV a1 VT Ns dV a1 VT Ns a2 VT Ns dV
( ) ( )
is being used to extract IL and Is. At the short circuit and the maximum IRs + V 1 dI
power points of the I–V characteristics, the current equation based on × exp − Rs + 1
a2 VT Ns Rsh dV
the SDM is written as follows [20]: (22)
[ ( ) ]
Isc = IL − Is exp
Isc Rs
− 1 − Gsh Isc Rs (13) At the maximum power point (MPP), the derivative of the power
aVT Ns equation with respect to the voltage (V) is equal to zero.
[ ( ) ] ⃒
( ) dP ⃒⃒
Imp = IL − Is exp
Imp Rs + Vmp
− 1 − Gsh Imp Rs + Vmp (14) =0 (23)
aVT Ns dV⃒(P=Pmp )

To simplify the equations’ writing, the following notations are However, in order to calculate the derivative of the output power
adopted. with respect to the output voltage (V) using the Eq. (22), it is necessary
( ) ( ) to employ the equation (24) below, which gives the expression of P.
Isc Rs Imp Rs + Vmp
Csc = exp − 1 ; Cmp = exp − 1;
aVT Ns aVT Ns P=I × V (24)
( ) (15)
Voc
Coc = exp − 1 Thus, the derivative of the power expression with respect to the
aVT Ns
voltage (V) can be found by deriving the Eq. (24) with respect to V,
The two equations’ system is obtained by combining the Eq. (13) and leading to the resulting Eq. (25):
the Eq. (14). dP dI
⎧ =V × +I (25)
⎪ Isc Rs dV dV
⎪
⎪ IL − Csc Is = Isc +
⎨ Rsh By combining the Eq. (23) and the Eq. (25), the current’s derivative
( ) (16)
⎪
⎪ Imp Rs + Vmp equation with respect to the voltage evaluated at the MPP can be ob­
⎪
⎩ IL − Cmp Is = Imp +
Rsh tained as given in the Eq. (26) below:
⃒
The system of Eq. (16) above is transformed into the matrix equation dI ⃒⃒ Imp
=− (26)
below: dV⃒(I=Imp ,V=Vmp ) Vmp

ASDM × λSDM = βSDM (17) By evaluating the Eq. (22) at the MPP and using the expression of the
Eq. (26), the Eq. (27) below can be obtained. The value of Rs can be then
where A is the two array matrix defined as: determined by solving this equation using the Brent method.
( ) ( ) ( )
1 − Csc
ASDM = (18) − Imp
+
Is1
1−
Imp Rs ( )
× Cmp,1 − 1 +
Is2
1−
Imp Rs
1 − Cmp
Vmp a1 VT Ns Vmp a2 VT Ns Vmp
( ) (27)
( ) 1 Imp Rs
and β and λ are two vectors defined as: × Cmp,2 − 1 + × 1− =0
Rsh Vmp
⎛ ⎞
Isc Rs
⎜ Isc + Rsh ⎟ Rsh is still the only unknown parameter in Eq. (27) to be determined
⎜ ⎟ before solving the equation. For the purpose of extracting some of the
βSDM = ⎜ ( )⎟ (19)
⎝ Imp Rs + Vmp ⎠
Imp + unknown parameters, many authors propose relying only on iterations
Rsh and a stopping criterion [35,39]. The technique is commonly used in the
( ) literature to get the Rs values because it is the easiest parameter to be
IL (Rs ) initialized as it takes small values, so near to zero [28,36,37]. However,
λSDM = (20)
Is (Rs )
for some PV modules, the series resistance takes values near 2, in this
To get the values of the two parameters IL and Is the Eq. (21) is used. case, the wrong choice of the increment steps can lead to the require­
ment of a large number of iterations, and that impacts the convergence
1
λSDM = A−SDM × βSDM (21) time. In addition to Rs, several authors proceed likewise to get the values
of the ideality factors, as well [35,39]. The last technique further
threatens the appropriateness of the execution time [6]. Therefore, to
3.2. Numerical methodology to extract Rs and Rsh reduce the number of iterations and guarantees the reproduction of the
algorithm, the increment steps must be well-determined and adapted to
In this section, the steps to be followed to extract the shunt resistance each PV generator of any technology. To that end, it is not obvious to
and the series resistance’s values based on the DDM and the SDM using find an increment step’s value for Rs because of its small values and its
only the key-points of the current-voltage characteristic are introduced. appearance in both exponential terms of the current equation of the
To that end, the extraction strategy is divided into two parts. The first is DDM in addition to its influence on the third term. Hence, the three I–V
the calculation of Rsh using an iterative process. The second is the use of curve’s zones are affected by this last sensitive parameter. As a solution,
the extracted value of Rsh to get the Rs value by solving a nonlinear and also to avoid the probable negative values of Rsh, a new technic
equation using Brent’s hybrid root-finding method [45]. based on the adjustment of the shunt resistance’s value is proposed in
the current work. Indeed, the method consists of choosing the research
3.2.1. Double-diode model interval of Rsh around its initial value extracted by calculating the
This subsection is dedicated to introducing the proposed technique slope’s value at the vicinity of the short circuit point or the theoretical
for extracting Rsh and Rs values in the context of DDM. The technique calculation of the initial value of Rsh. The principle consists in solving
involves solving the derivative equation of the current at the maximum the equation (28) below:
power point (MPP) using the Brent method. To do this, the derivative of ⃒
dI ⃒⃒ 1
the current’s equation of Eq. (2) is computed with respect to the voltage =− (28)
dV⃒I=Isc,V=0 Rsh0
V, resulting in the following expression [17]:

5
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

Fig. 3. Flow chart for the DDM.

The supplementary benefit of the proposed technique for Rsh is its The flow chart for the SDM is given in Fig. 4.
flexibility so that it can be adapted to the case of data availability, in this
case, the Eq. (28) can be solved graphically. Otherwise, the theoretical 3.3. Estimating the performance of PV modules at various weather
initial value of Rsh given below can be used [36]: conditions
Vmp Voc − Vmp
Rsh0 = − (29) Calculating the electrical parameters of PV generators at the stan­
Isc − Imp Imp
dard test conditions (T = 25 ◦ C, G = 1 kW/m2, and AM = 1.5) is
The searching interval is defined with consideration of the visible 5% considered mandatory to achieve the estimation of PV modules’ per­
of error around the initial value, to ensure selecting another initial value formance when operating under external conditions. However, the
of Rsh, which corresponds to a slope below the desired slope that will calculated parameters at STC are affected by weather variations,
match the most effective value of Rsh. Then, the stopping criterion including temperature and irradiance. Thus, the transfer of parameters’
consists in minimizing the absolute error between the calculated MPP values from STC to non-STC is a needed task. Accordingly, the transfer
value (PC,max) and the measured one (PM,max) given in Eq. (30), until the equations depending on T and G levels are widely used in the literature
desired precision corresponding to the tolerance (ε) is reached. to meet that need [6,35,37].
⃒
⃒ The ideality factors of the DDM are presumed unaffected neither by
AE(W) = |PC,max − PM,max ⃒ (30) the irradiance nor temperature variations [6,35].

The flow chart of the proposed method based on the DDM is given in a1 (T, G) = a1,ref (32)
Fig. 3.
a2 (T, G) = a2,ref (33)
3.2.2. Single-diode model The shunt resistance and the series resistance’s values vary only
To make the numerical technique adapted with the SDM to extract when the level of irradiance varies, and they are linked to the G level by
the value of Rs, the Eq. (31) obtained from the calculation of the current the following formulas [6]:
equation’s derivative with respect to the voltage (V) at the MPP [17,47],
is solved using Brent’s method. Rsh,ref (T, G) = Rsh,ref
Gref
(34)
( ) ( ) G
− Imp Is Imp Rs ( ) 1 Imp Rs
+ 1− × Cmp − 1 + × 1− =0 (31)
Vmp aVT Ns Vmp Rsh Vmp Gref
Rs (T, G) = Rs,ref (35)
G
The value of Rsh is extracted by following the same numerical
method proposed for the DDM.

6
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

Fig. 4. Flow chart for the SDM.

The values taken by Is1 and Is2 are influenced by the temperature represent the widely installed technologies such as poly-crystalline,
variations and can be determined at non-STC using Is1,ref and Is2,ref mono-crystalline, and thin film technologies. Among them, mono-
values calculated at the STC, the equations allowing this transformation crystalline PV systems are considered the most efficient and are well-
are given below [35,37]: suited for warm and sunny locations. While polycrystalline modules
( ( )) ( )3 may have slightly lower performance than mono-crystalline ones, they
Is1 (T, G) = Is1,ref × exp
T Eg,ref Eg
− ×
T
(36) offer a competitive cost-efficiency ratio [48]. On the other hand, as re­
VT a1,ref Tref T Tref ported by the authors in Ref. [49], thin film technology is characterized
( ( )) ( )3 by a lower temperature coefficients, maintaining stable performance
Is2 (T, G) = Is2,ref × exp
T Eg,ref Eg
− ×
T
(37) even at higher temperatures. Accordingly, this technology is charac­
VT a2,ref Tref T Tref terized by high performance ratio. Thus, thin film modules represent an
excellent choice for hotter climates, despite their lower efficiency [48,
where Eg notation represents the band gap energy at external conditions 49].
defined as [31,35]. In this work, to comprehensively and meaningfully evaluate the new
( (
Eg = Eg,ref 1 + 0.0002677 × Tref − T
))
(38) approach, the validation of the modeling technique includes six widely
used PV modules from previous technologies [6,32,37]. The mean
The light-generated current is affected by both temperature and points’ values at Standard Test Conditions (STC) corresponding to each
irradiance variations according to the following equation: PV module and their required temperature coefficients are provided in
( (
(
)) G
) Table 1.
Iph (T, G) = Iph,ref 1 − KIsc Tref − T (39) For a fair comparison, at first, the electrical parameters of the
Gref
equivalent circuit models are calculated under STC (G = 1 kW/m2, T =
The index “ref” is used to designate the parameter’s values at the 25 ◦ C, and AM = 1.5) and then compared with other proposed ap­
reference conditions (STC). proaches in the literature. Next, the I–V characteristics are estimated
based on the extracted parameters and tested against the measured data
4. Results and discussion of the PV modules using some statistical indicators. The resulting values
of this last comparison are also compared with the results provided by
This section is dedicated to validating the effectiveness of the pro­ some of the most reviewed approaches in the literature. During the
posed method. In the literature, numerous PV modules have been second phase, the achieved results at STC are used to prediction at non-
extensively analyzed in the PV modeling field. Generally, these modules STC.

7
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

Table 1
PV modules’ specific parameters.
Poly-crystalline Mono-crystalline Thin film

BP solar MSX-60 KD245GH-4FB2 S75 SQ150-PC SM55 ST40

Ns 36 60 36 72 36 36
Isc (A) 3.8 8.91 4.7 4.80 3.45 2.68
Imp(A) 3.5 8.23 4.26 4.40 3.15 2.41
Vmp(V) 17.1 29.8 17.6 34 17.4 16.6
Voc(V) 21.1 36.9 21.6 43.4 21.7 23.3
3 3 3 3 3 4
KIsc (A/◦ C) 3 × 10− 5.35 × 10− 2 × 10− 1.4 × 10− 1.4 × 10− 3.5 × 10−

4.1. Effect of the shunt resistance’s adjustment on the I–V curves’ its least value it implies the highest slope near Isc. Then, when Rsh in­
prediction and results for STC creases, the inclination decreases to reach its final value that matches
the MPP and then, fits the linear zone with great precision. The previous
The current work introduces several new extraction techniques. observation is enhanced by Fig. 5(b) giving the individual absolute er­
Thus, this subsection is laid aside first for the visualization of the shunt rors calculated between the measured data and the calculated charac­
resistance’s adjustment effect on the estimation of the PV generators’ teristic for each Rsh value. It can be seen that by adjusting Rsh, the
characteristics. Second, to test the impact of eliminating the two pa­ absolute error minimizes for the great majority of the predicted curve.
rameters related to the second diode of the DDM to return to the SDM on Table 2 gives the calculated values of the DDM parameters using the
the proposed method’s performance. The third task to which this sub­ slope adjustment technique in comparison with three of the widely
section is devoted is the test of the proposed approach’s effectiveness at reviewed works given in Refs. [17,37], and [50]. The root mean square
the STC by validating the accuracy against the methods introduced in error (RMSE) is chosen as the statistical indicator to measure the
the literature. And then finally, to visualize the predicted I–V curves at resemblance ratio between the measured and calculated characteristics
the reference conditions. based on the extracted DDM parameters. This choice also enables an
accurate comparison with previous techniques. According to the results,
4.1.1. Effect of shunt resistance’s adjustment the new modeling approach demonstrates the best similitude to the
The test in this part is done over the multi-crystalline PV panel BP measured data, with an RMSE value of 0.0343 A, which is the least value
solar MSX-60 operating at STC. Fig. 5 provides a clear visualization of among the other methods. The method in Ref. [17] is a purely numerical
the effect of adjusting the Rsh value on the slope adjustment at the short method based on a resolution of seven equations using the ‘fsolve’
circuit point and also its effect on the slope of the I–V curve’s first zone function of MATLAB software. However, the technique is based on
for the BP solar MSX-60 PV module. As Fig. 5(a) shows, when Rsh takes several approximations to simplify the tedious calculation, in addition to

Fig. 5. (a) Estimated characteristic for MSX-60 PV module for various Rsh values compared to the measured characteristic. (b) Absolute errors matching each
Rsh value.

Table 2
Extracted DDM’s parameters for the MSX-60 PV module at STC using the Rsh adjustment technic compared to three other modeling methods of literature.
BP solar MSX-60

IL(A) a1 a2 Is1(A) Is2(A) Rsh(Ω) Rs(Ω) RMSE (A)

Kumar et al., [17] 3.8046 0.99859 2.0014 3.9901 × 10− 10 4.033 × 10− 06 280.2171 0.3397 0.1908
Zhuo et al., [37] 3.808361 1 2 4.597752 × 10− 10 4.597752 × 10− 10 169.081308 0.372046 0.1666
Elbaset et al., [50] 3.8084 1.0003 1.9997 4.8723 × 10− 10 6.1528 × 10− 10 169.0471 0.3692 0.1379
Proposed DDM Method 3.8618 1 2 6.0842 × 10− 10 3.1092 × 10− 06 286.1209 0.2904 0.0343

8
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

the use of seven initializations. These factors have led to the least pre­ applied for the DDM by providing 0.0334 A as RMSE value for S75 PV
cision of estimation by providing 0.1908 A as an RMSE value, which is module and 0.0364 A for the SM55 PV module. Moreover, when the
the highest value. Moreover, it can be seen that the values of the two adjusting technique is applied to the SDM, it provides high levels of
ideality factors consistently return toward 1 for a1 and 2 for a2, this precision compared with the existing works in the literature. If
observation is also confirmed by the results provided by the works in comparing the obtained DDM’s parameters with the SDM’s ones, it can
Ref. [50]. The last observation trend aligns with the physical signifi­ be observed that for both PV modules, the SDM provides almost the
cance of these two parameters. same values of IL, Is1, a1, and Rs as those yielded by the DDM. However, it
can be observed from the difference between the achieved Rsh values
4.1.2. Predicting PV modules performance at STC using the modeling DDM that when the approach is used for the SDM, it converges relatively
and SDM methods faster than the DDM, as the SDM gives the lowest values of this
Intending to test the effectiveness of the new modeling approach parameter. However, since Rsh takes high values and is incremented
applied for both the DDM and the SDM against the existing numerical based on a percentage technique, the convergence time is not heavily
methods in the literature, the multi-crystalline S75 and the mono- influenced by the small differences in Rsh values. Furthermore, the small
crystalline SM55 PV panels operating at reference conditions are used difference between the values of Rsh of the two models and the consid­
in this part. Table 3 compares the DDM’s parameters and SDM’s five eration of the existence of the recombination phenomenon give the DDM
parameters extracted using the proposed modeling technique applied to the preference at the precision level. So it can be concluded that even if
the multi-crystalline S75 PV panel, also compared to six other works the DDM requires more steps to converge, it achieves the most accurate
introduced in the literature. Table 4 gives the same study done over the results. The absolute neglect of the resistive effects inside PV generators
mono-crystalline SM55 PV. For a relevant evaluation, the new modeling by the authors in Ref. [26] is the reason for obtaining less accurate
DDM method is also tested over the multi-crystalline KD245GH-4FB2, perdition by this technique for both PV modules. While, the high values
the mono-crystalline SQ150-PC PV module, and the thin-film ST40 PV of RMSE obtained by the methods in Refs. [32,33] for the two PV panels
module, and its efficiency is compared with three different methods of are explained by the use of a series of approximations including IL=Isc
literature. The results are shown, respectively, in Tables 5–7. and Is1 = Is2 equalities that have no physical explanation. The method in
As appears in Tables 3 and 4 for the multi-crystalline S75 and the Ref. [29] is based on a meta-heuristic technique affected by the addi­
mono-crystalline SM55 the highest estimation’s accuracy among the tional parameters related to the used algorithm, and that influences as
other predicting techniques is offered by the proposed modeling method well the estimation’s precision. The mean disadvantage of the works in

Table 3
Extracted DDM and SDM’s parameters for the multi-crystalline S75 PV module at STC using the adjustment technique compared to three other modeling methods of
literature.
Shell S75

IL(A) a1 a2 Is1(A) Is2(A) Rsh(Ω) Rs(Ω) RMSE (A)

− 09 − 06
Tifidat et al., [6] 4.70163 1.09371 1.9420 2.114 × 10 4.24173 × 10 756.458 0.2612 0.075788
Chitti Babu et al., [26] 4.7 1.82 3.36 1.23 × 10− 05 3.18 × 10− 05 ∞ 0 0.222448
Yahya-Khotbehsara et al., [28] 4.7 1.14 2.6 5.68 × 10− 09 1.47 × 10− 08 103.58 0.23 0.121921
Ishaque et al., [32] 4.7 1 1.3 3.39 × 10− 10 3.39 × 10− 10 84.38 0.27 0.089657
Khezzar et al., [33] 4.7 1 1.3 3.388 × 10− 10 3.388 × 10− 10 78.641 0.2541 0.107319
Chin et al., [29] 4.7 1.14 2.60 1.63 × 10− 08 9.29 × 10− 06 403.16 0.25 1.147528
Proposed Method for SDM 4.70987 1 – 3.2976 × 10− 10 – 159.011 0.33399 0.038862
Proposed Method for DDM 4.70936 1 2 3.3026 × 10− 10 1.4034 × 10− 08 170.507 0.33975 0.033388

Table 4
Extracted DDM and SDM’s parameters for the Mono-crystalline SM55 PV module at STC using the adjustment technique compared to three other modeling methods of
literature.
Shell SM55

IL(A) a1 a2 Is1(A) Is2(A) Rsh(Ω) Rs(Ω) RMSE(A)

05 05
Chitti Babu et al., [26] 3.45 1.89 3.65 1.4300 × 10− 3.7000 × 10− ∞ 0 0.1062
08 08
Yahya-Khotbehsara et al., [28] 3.45 1.26 2.84 2.7500 × 10− 7.1100 × 10− 233.4600 0.36000 0.0789
10 10
Khezzar et al., [33] 3.45 1 1.3 2.4873 × 10− 2.4873 × 10− 93.03299 0.29890 0.1211
10 10
Lun et al., [34] 3.4619 1.01658 1.01658 1.5529 × 10− 1.5524 × 10− 144.2891 0.49696 0.0577
10 06
Chennoufi et al., [35] 3.4604 0.98580 2.01420 1.4249 × 10− 2.1336 × 10− 151.3306 0.45670 0.0642
10 10
Zhuo et al., [37] 3.4624 1 2 2.1593 × 10− 2.1593 × 10− 141.2190 0.50639 0.0478
10
Proposed Method for SDM 3.4621 1 – 2.3781 × 10− – 132.1152 0.46150 0.0393
10 08
Proposed Method for DDM 3.4616 1 2 2.3812 × 10− 4.1270 × 10− 140.3952 0.47230 0.0364

Table 5
Extracted DDM’s parameters for the multi-crystalline KD245GH-4FB2 PV module at STC using the adjustment technique compared to three other modeling methods of
literature.
Keyocera KD245GH-4FB2

IL(A) a1 a2 Is1(A) Is2(A) Rsh(Ω) Rs(Ω) RMSE (A)

10
Enebish et al., [51] 8.9335 1.0004 2.0008 3.5748 × 10− − 1.1878 × 10− 6 120.1507 0.3163 0.0475
10
Kumar et al., [17] 8.9202 0.9954 2.0044 2.7637 × 10− 6.4751 × 10− 6 247.9658 0.2834 0.07356
10
Orioli et al., [39] 8.9337 0.9735 1.5649 1.8084 × 10− 1.9062 × 10− 10 120.1628 0.3192 0.04855
10
Proposed Method for DDM 8.9310 1 2 3.4743 × 10− 6.6148 × 10− 8 131.5809 0.3103 0.0444

9
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

Table 6
Extracted DDM’s parameters for the mono-crystalline SQ150-PC PV module at STC using the adjustment technique compared to three other modeling methods of
literature.
Shell SQ150-PC

IL(A) a1 a2 Is1(A) Is2(A) Rsh(Ω) Rs(Ω) RMSE (A)

10 10
Khezzar et al., [33] 4.8 1 1.3 3.1068 × 10− 3.1068 × 10− 275 0.9 0.04299
05 04
Chitti Babu et al., [26] 4.8 2.0300 ∞ 4.5836 × 10− 1.1875 × 10− ∞ 0 0.27081
19 07
Orioli et al., [39] 4.8054 0.73076 1.47759 1.8765 × 10− 5.9857 × 10− 444.1415 0.4981 0.12066
10 06
Proposed DDM Method 4.8092 1 2 2.8725 × 10− 4.1830 × 10− 444.6400 0.8503 0.0294

Table 7
Extracted DDM’s parameters for the thin-film ST40 PV module at STC using the adjustment technique compared to three other modeling methods of literature.
Shell ST40

IL(A) a1 a2 Is1(A) Is2(A) Rsh(Ω) Rs(Ω) RMSE (A)

Yahya-Khotbehsara et al., [28] 2.68 1.52 2.15 1.65 × 10− 07 4.28 × 10− 07 1350.30 1.40 0.06393
Ishaque et al., [32] 2.68 1 1.3 3.07 × 10− 11 3.07 × 10− 11 198.94 1.71 0.0957
Zhuo et al., [37] 2.70773 1 2 2.9755 × 10− 11 2.97551 × 10− 11 168.9173 1.74749 0.094321
Proposed DDM Method 2.7115 1 2 2.2134 × 10− 11 1.2153 × 10− 06 127.6626 1.5184 0.0319

Ref. [34] that may explain the obtained RMSE value for the SM55 PV the STC. The predicted curves are presented by the continuous red line,
module is the equality between the two ideality factors and the two the methods of literature are given by the colored lines, and the
saturation currents’ equality. measured current-voltage curves are displayed using black markers. For
According to the results in Table 5, the first observation to be made is both PV modules, it can be seen that in the vicinity of the MPP, the
that the proposed adjustment technique yields the lowest value of RMSE introduced technique in the current paper ensures high prediction ac­
for KD245GH-4FB2 PV when compared to other methods of the litera­ curacy, and the estimated characteristics match with high precision the
ture. The second note is that although the works in Refs. [17,51] use two experimental data. Figs. 6(b) and 7(b) give the individual absolute errors
different techniques, they both converge toward 1 for a1 and 2 for a2. matching each of the used methods for the extraction of I–V character­
The results of Table 6 confirm the effectiveness of the proposed method istics of the two PV modules S75 and SM55, respectively. The calculated
which yields the least values of the selected statistical indicators also for absolute errors at the vicinity of the MPP enhance the previous obser­
the mono-crystalline SQ150-PC PV module. Furthermore, the used ap­ vation on the proposed method’s effectiveness for the MPP’s prediction.
proximations in Ref. [26] have supplied a physical unexplained value of With the aim of testing the speed of convergence of the proposed
the second ideality factor for the SQ150-PC, and that has also led to an method, this last, as well as the three methods in Refs. [26,28], and [32],
increase of the corresponding RMSE value of this method. The same were implemented in MATLAB R2012ab software and have been run
observation concerning the method’s accuracy can be noted from several times on an hp 8470p EliteBook with 4 GB RAM and an Intel
Table 7 for the thin-film ST40 PV module. The proposed methods in Core i5-3320M CPU, 2.60 GHz processor. The provided computational
Refs. [32,37] assume the equality of the two reverse saturation currents time and errors at the MPP for S75 and SM55 PV modules are shown in
and that has strongly affected the efficiency of the prediction. Table 8. As reported in the table, the best estimation of MPP is provided
Figs. 6(a) and 7(a) show the estimated I–V characteristics using the by the proposed adjustment technique at the least execution time for
proposed adjustment approach applied to the DDM tested over the two both PV modules compared to the other methods. These two last results
PV modules; the shell S75 and the shell SM55, respectively, operating at are achieved first due to the use of absolute error minimization at the

Fig. 6. (a) Estimated characteristic for the multi-crystalline S75 PV module at STC compared with the measured data and other methods of the literature. (b) In­
dividual absolute errors matching each modeling method.

10
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

Fig. 7. (a) Estimated characteristic for the mono-crystalline SM55 PV module at STC compared with the measured data and other methods of the literature. (b)
Individual absolute errors matching each modeling method.

Table 8
Comparison of the proposed model’s convergence time and accuracy for MPP’s prediction with existing modeling techniques.
Chitti Babu et al., [26] Yahya-Khotbehsara et al., [28] Ishaque et al., [32] Proposed DDM Method

Poly-crystalline KD245GH-4FB2 PM,max (W) 245.2540 245.2540 245.2540 245.2540

PC,max (W) 245.2483 245.2533 245.2540 245.2540
06 07
AEMpp 0.0057 0.0007 3.9999 × 10− 7.136397 × 10−
t (s) 0.3164 0.1057 0.0910 0.083505

Mono-crystalline SM55 PM,max (W) 54.8100 54.8100 54.8100 54.8100

PC,max (W) 54.6876 54.7885 54.8129 54.8100
06
AEMpp 0.1224 0.0215 0.0029 1.0869 × 10−
t (s) 1.3022 0.2213 0.0887 0.074438

Fig. 8. (a) Estimated characteristic for the KD245GH-4FB2 PV module operating under various levels of illumination and standard T compared with the measured
data and other methods of the literature. (b) Estimated and measured curves for the KD245GH-4FB2 PV module operating under various temperatures and stan­
dard G.

11
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

MPP and second due to the exploitation of the new technique allowing reported in the literature. This study is done over three PV modules, The
the fast extraction of the shunt resistance based on the increment using multi-crystalline KD245GH-4FB2, the mono-crystalline SQ150-PC, and
the percentage method. By contrast, the other methods of literature are the thin-film ST40 PV panels. Fig. 8(a), Fig. 9(a), and Fig. 10(a) show the
based on random increments of Rs using randomly selected steps chosen calculated and measured curves under the reference temperature and
by the authors and applied for all PV modules, the fact which explains various G levels for the three PV modules, respectively. While, Figs. 8(b),
the high execution times of these methods. 9(b) and 10(b) give the calculated and measured curves under the
reference irradiance and various T levels for the same PV generators.
4.2. Predicting PV modules performance at non-STC using the modeling From the figures, as the calculated curves highly match the measured
DDM method data for all PV generators used for the test, it can be concluded that the
new technique allows the estimation of PV performance under different
For the purpose of evaluating the performance of the modeling weather conditions with great precision.
technique at non-STC, the DDM’s parameters are utilized to estimate the The modeling methods’ RMSE related to each operating value of G
PV generators’ performance at external conditions by transferring their and T = 25 ◦ C calculated for the KD245GH-4FB2, the SQ150-PC, and the
values to non-STC and calculating the I–V characteristics. Subsequently, ST40 PV panels are given in Fig. 11(a), Fig. 12(a), and Fig. 13(a),
the correspondence of the predicted performance is tested against the respectively. It can be noted from the three figures that the new
measured I–V curves, and the results are compared to other methods adjustment technique offers a high level of precision and follows the

Fig. 9. (a) Estimated characteristic for the SQ150-PC PV module operating under various levels of illumination and standard T compared with the measured data and
other methods of the literature. (b) Estimated and measured curves for the SQ150-PC PV module operating under various temperatures and standard G.

Fig. 10. (a) Estimated characteristic for the ST40 PV module operating under various levels of illumination and standard T compared with the measured data and
other methods of literature. (b) Estimated and measured curves for the ST40 PV module operating under various temperatures and standard G.

12
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

Fig. 11. (a) RMSE of each modeling method applied for the KD245GH-4FB2 module operating under various levels of illumination and standard T. (b) RMSE of each
modeling method applied for the KD245GH-4FB2 PV module operating under various temperatures and standard G.

Fig. 12. (a) RMSE of each modeling method applied for the SQ150-PC PV module operating under various levels of illumination and standard T. (b) RMSE of each
modeling method applied for the SQ150-PC PV module operating under various temperatures and standard G.

illumination’s variations with great accuracy, and that is concluded 5. Conclusion

from the RMSE that does not exceed 0.104 A for the multi-crystalline,
0.0411 A for the mono-crystalline, and 0.0466 A for the thin-film. On A fast modeling technique to calculate the parameters of the two-
the other side, for the three technologies, respectively, the works in diode model of PV modules based on the slope adjustment using only
Ref. [39] reach 0.169 A, while 0.364 A is achieved by the method in the key-points’ values at STC is introduced in the current paper. The
Ref. [26] and 0.132 A by the technique in Ref. [32]. Figs. 11(b), 12(b) method consists of extracting the three parameters; IL, Is1, and Is2 using
and 13(b) show the methods’ RMSE related to each operating value of T the analytical resolution of a three equation’s system. The series resis­
and G = 1 kW/m2, calculated for the same PV modules, respectively. The tance is extracted using an accurate numerical resolution based on the
proposed parameter extraction technique provides in the worst cases Brent’s solver algorithm, and the shunt resistance is calculated based on
RMSE = 0.13 A for the multi-crystalline, RMSE = 0.0542 A for the a new fast adjusting technique. The modeling approach was also
mono-crystalline, and RMSE = 0.0505 A for the thin-film, which are the adapted to the single-diode model. The accuracy and the speed of
least values provided when compared with the available methods in the convergence are evaluated against a considerable number of the avail­
literature. Consequently, it can be noted that the method in this work able methods in the literature using six modules of various PV tech­
follows the weather variation with great accuracy when predicting I–V nologies when operating STC and non-STC conditions. The test of the
characteristics. efficiency was done using some statistical indicators for which the
proposed technique gives the best values in the least execution time,

13
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

Fig. 13. (a) RMSE of each modeling method applied for the ST40 PV module operating under various levels of illumination and standard T. (b) RMSE of each
modeling method applied for the ST40 PV module operating under various temperatures and standard G.

which makes the new method applicable for performance assessment at and adaptive damping method of Berndt-Hall-Hall-Hausman”, Sol. Energy 243
(2022) 35–61, https://doi.org/10.1016/j.solener.2022.07.029.
real-time and also for MPP tracking applications.
[9] J.D. Bastidas-Rodriguez, G. Petrone, C.A. Ramos-Paja, G. Spagnuolo, A genetic
algorithm for identifying the single diode model parameters of a photovoltaic
CRediT authorship contribution statement panel, Math. Comput. Simulat. (2015), https://doi.org/10.1016/j.
matcom.2015.10.008.
[10] G. Xiong, J. Zhang, D. Shi, Y. He, Parameter extraction of solar photovoltaic models
Kawtar Tifidat: Conceptualization, Methodology, Formal analysis, using an improved whale optimization algorithm, Energy Convers. Manag. 174
Software, Validation, Visualization, Investigation, Writing – original (2018) 388–405, https://doi.org/10.1016/j.enconman.2018.08.053.
draft, Writing – review & editing. Noureddine Maouhoub: Conceptu­ [11] K. Tifidat, N. Maouhoub, F.E. Ait Salah, Modeling approach for extracting the
single-diode model parameters and predicting PV modules performance, AIP Conf.
alization, Methodology, Validation, Investigation, Supervision, Writing Proc. 2814 (1) (2023), 040006, https://doi.org/10.1063/5.0148726, 11 July
– review & editing. 2023.
[12] R.M. Rizk-Allah, A.A. El-Fergany, Conscious neighborhood scheme-based
Laplacian barnacles mating algorithm for parameters optimization of photovoltaic
Declaration of competing interest single- and double-diode models, Energy Convers. Manag. 226 (2020), 113522,
https://doi.org/10.1016/j.enconman.2020.113522.
[13] A. Laudani, F. Rigatani Fulginei, A. Salvini, “High performing extraction procedure
The authors declare that they have no known competing financial for the one-diode model of a photovoltaic panel from experimental I–V curves by
interests or personal relationships that could have appeared to influence using reduced forms”, Sol. Energy 103 (2014) 316–326, https://doi.org/10.1016/j.
the work reported in this paper. solener.2014.02.014.
[14] P. Hao, Y. Zhang, H. Lu, Z. Lang, A novel method for parameter identification and
performance estimation of PV module under varying operating conditions, Energy
References Convers. Manag. (2021), https://doi.org/10.1016/j.enconman.2021.114689.
[15] M.R. Arabshahi, H. Torkamana, A. Keyhanib, A method for hybrid extraction of
[1] M. Zaimi, H. El Achouby, O. Zegoudi, A. Ibral, E.M. Assaid, “Numerical Method and single-diode model parameters of photovoltaics, Renew. Energy S0960–1481 (20)
New Analytical Models for Determining Temporal Changes of Model-Parameters to (2020), https://doi.org/10.1016/j.renene.2020.05.035, 30733-3.
Predict Maximum Power and Efficiency of PV Module Operating Outdoor under [16] S. Bana, R. Saini, A mathematical modeling framework to evaluate the
Arbitrary Conditions”, vol. 177, ELSEVIER. Energy Conversion and Management, performance of single diode and double diode based SPV systems, Energy Rep. 2
2020, pp. 258–271, https://doi.org/10.1016/j.enconman.2018.09.054. (2016) 171–187, https://doi.org/10.1016/j.egyr.2016.06.004.
[2] M. Fathi, J.A. Parian, “Intelligent MPPT for Photovoltaic Panels Using a Novel [17] M. Kumar, A. Kumar, An efficient parameters extraction technique of photovoltaic
Fuzzy Logic and Artificial Neural Networks Based on Evolutionary Algorithms”, models for performance assessment, Sol. Energy (2017) 192–206, https://doi.org/
vol. 7, ELSEVIER. Energy Reports, 2021, pp. 1338–1348, https://doi.org/10.1016/ 10.1016/j.solener.2017.09.046.
j.egyr.2021.02.051. [18] N. Maouhoub, Analytical identification method for the single diode model
[3] H. El Achouby, M. Zaimi, A. Ibral, E.M. Assaid, “New analytical approach for parameters of a photovoltaic panel using datasheet values, J. Nano- Electron. Phys.
modelling effects of temperature and irradiance on physical parameters of 9 (6) (2017), 06011, https://doi.org/10.21272/jnep.9(6).06011.
photovoltaic solar module”, Energy Convers. Manag. 177 (2018) 258–271, https:// [19] H. Ibrahim, N. Anani, Evaluation of analytical methods for parameter extraction of
doi.org/10.1016/j.enconman.2018.09.054. PV modules, Energy Proc. 134 (2017) 69–78, https://doi.org/10.1016/j.
[4] K. Tifidat, N. Maouhoub, A. Benahmida, A new reduced form for real-time egypro.2017.09.601.
identification of PV panels operating under arbitrary conditions, Int. J. Energy [20] K. Tifidat, N. Maouhoub, S. Askar, M. Abouhawwash, Numerical procedure for
Optim. Eng. 11 (2) (2022) 1–23, https://doi.org/10.4018/IJEOE.309415. accurate simulation of photovoltaic modules performance based on the
[5] A.H. Eisapour, M. Eisapour, M.J. Hosseini, A.H. Shafaghat, P.T. Sardari, A. identification of the single-diode model parameters, Energy Rep. 9 (2023)
A. Ranjbar, “Toward a highly efficient photovoltaic thermal module: energy and 5532–5544, https://doi.org/10.1016/j.egyr.2023.04.378.
exergy analysis”, Renew. Energy 169 (2021) 1351–1372, https://doi.org/10.1016/ [21] M. Qaraad, S. Amjad, N.K. Hussein, M. Badawy, S. Mirjalili, M.A. Elhosseini,
j.renene.2021.01.110. Photovoltaic parameter estimation using improved moth flame algorithms with
[6] K. Tifidat, N. Maouhoub, A. Benahmida, F.E. Ait Salah, An Accurate Approach for local escape operators, Comput. Electr. Eng. 106 (2023), 108603, https://doi.org/
Modeling I-V Characteristics of Photovoltaic Generators Based on the Two-Diode 10.1016/j.compeleceng.2023.108603.
Model”, vol. X, Energy Conversion and Management, 2022, https://doi.org/ [22] N. Boutana, A. Mellit, S. Haddad, A. Rabhi, A. Massi Pavan, An explicit I-V model
10.1016/j.ecmx.2022.100205. for photovoltaic module technologies, Energy Convers. Manag. 138 (2017)
[7] A. Elkholy, A.A. Abou El-Ela, Optimal parameters estimation and modelling of 400–412, https://doi.org/10.1016/j.enconman.2017.02.016.
photovoltaic modules using analytical method, Heliyon 5 (2019), e02137, https:// [23] A. Benahmida, N. Maouhoub, K. Tifidat, H. Sahsah, Extraction of photovoltaic
doi.org/10.1016/j.heliyon.2019.e02137. generator parameters through combination of an analytical and iterative approach,
[8] H.M. Ridha, H. Hizam, S. Mirjalili, M.L. Othman, M. Effendy Ya’acob, Int. J. Electr. Comput. Eng. 12 (No. 5) (2022) 4529–4537, https://doi.org/
M. Ahmadipour, “Novel parameter extraction for Single, Double, and three diodes 10.11591/ijece.v12i5.pp4529-4537.
photovoltaic models based on robust adaptive arithmetic optimization algorithm

14
K. Tifidat and N. Maouhoub Renewable Energy 216 (2023) 119102

[24] S. Li, W. Gong, X. Yan, C. Hua, D. Bai, L. Wang, L. Gao, Parameter extraction of [38] D.S.H. Chan, J.C.H. Phang, Analytical methods for the extraction of solar-cell
photovoltaic models using an improved teaching learning-based optimization, single- and double-diode model parameters from I-V characteristics, IEEE Trans.
Energy Convers. Manag. 186 (2019) 293–305, https://doi.org/10.1016/j. Electron. Dev. 34 (2) (1987) 286–293, https://doi.org/10.1109/T-ED.1987.22920.
enconman.2019.02.048. [39] A. Orioli, A. Di Gangi, A procedure to evaluate the seven parameters of the two-
[25] C. Zhang, Y. Zhang, J. Su, T. Gu, M. Yang, Modeling and prediction of PV module diode model for photovoltaic modules, Renew. Energy (2019), https://doi.org/
performance under different operating conditions based on power-law I-V model, 10.1016/j.renene.2019.02.122.
IEEE J. Photovoltaics (2020), https://doi.org/10.1109/JPHOTOV.2020.3016607. [40] K. Yu, J.J. Liang, B.Y. Qu, X. Chen, H. Wang, Parameters identification of
[26] B. Chitti Babu, S. Gurjar, A novel simplified two-diode model of photovoltaic (PV) photovoltaic models using an improved JAYA optimization algorithm, Energy
module, IEEE J. Photovoltaics 4 (NO. 4) (2014), https://doi.org/10.1109/ Convers. Manag. 150 (2017) 742–753, https://doi.org/10.1016/j.
JPHOTOV.2014.2316371. enconman.2017.08.063.
[27] S. Gupta, H. Tiwari, Ma Fozdar, V. Chandna, “Development of a two diode model [41] K. Yu, J.J. Liang, B.Y. Qu, Z. Cheng, H. Wang, Multiple learning backtracking
for photovoltaic modules suitable for use in simulation studies”, Asia-Pac. Power search algorithm for estimating parameters of photovoltaic models, Appl. Energy
Energy Eng. Conf. (2012) 1–4, https://doi.org/10.1109/APPEEC.2012.630720. 226 (15) (2018) 408–422, https://doi.org/10.1016/j.apenergy.2018.06.010.
[28] A. Yahya-Khotbehsara, A. Shahhoseini, A fast modeling of the double-diode model [42] K.M. Sallam, M.A. Hossain, R.K. Chakrabortty, M.J. Ryan, An improved gaining-
for PV modules using combined analytical and numerical approach, Sol. Energy sharing knowledge algorithm for parameter extraction of photovoltaic models,
162 (2018) 403–409, https://doi.org/10.1016/j.solener.2018.01.047. Energy Convers. Manag. 237 (2021), 114030, https://doi.org/10.1016/j.
[29] V.J. Chin, Z. Salam, K. Ishaque, “An accurate modelling of the two-diode model of enconman.2021.114030.
PV module using a hybrid solution based on differential evolution”, Energy [43] M. Wang, X. Xu, Z. Yan, H. Wang, An online optimization method for extracting
Convers. Manag. 124 (2016) 42–50, https://doi.org/10.1016/j. parameters of multi-parameter PV module model based on adaptive Levenberg-
enconman.2016.06.076. Marquardt algorithm, Energy Convers. Manag. 245 (2021), 114611, https://doi.
[30] V.J. Chin, Z. Salam, K. Ishaque, An accurate two diode model computation for CIS org/10.1016/j.enconman.2021.114611.
thin film PV module using the hybrid approach, in: Poc. Electric Power and Energy [44] C. -t. Sah, R.N. Noyce, W. Shockley, Carrier generation and recombination in P-N
Conversion Systems (EPECS). Conf, IEEE., 2015, pp. 1–6, https://doi.org/10.1109/ junctions and P-N junction characteristics, Proc. IRE 45 (9) (1957) 1228–1243,
EPECS.2015.7368493. https://doi.org/10.1109/JRPROC.1957.278528. Sept. 1957.
[31] R. Majdoul, E. Abdelmounim, M. Aboulfatah, A.W. Touati, A. Moutabir, [45] R.P. Brent, Chapter 4: an algorithm with guaranteed convergence for finding a zero
A. Abouloifa, “Combined analytical and numerical approach to determine the four of a function, in: Algorithms for Minimization without Derivatives, Prentice-Hall,
parameters of the photovoltaic cells models”, in: IEEE 2015 International Englewood Cliffs, NJ, 1973. ISBN 0-13-022335-2.
Conference on Electrical and Information Technologies (ICEIT) - Marrakech, [46] D. Wang, X. Sun, H. Kang, Y. Shen, Q. Chen, Heterogeneous differential evolution
Morocco (2015.3.25-2015.3.27), 2015, https://doi.org/10.1109/ algorithm for parameter estimation of solar photovoltaic models, Energy Rep. 8
eitech.2015.7162977. (2022) 4724–4746, https://doi.org/10.1016/j.egyr.2022.03.144.
[32] K. Ishaque, Z. Salam, H. Taheri, Simple, fast and accurate two-diode model for [47] K. Tifidat, N. Maouhoub, A. Benahmida, An efficient numerical method and new
photovoltaic modules, Sol. Energy Mater. Sol. Cells 95 (2011) 586–594, https:// analytical model for the prediction of the five parameters of photovoltaic
doi.org/10.1016/j.solmat.2010.09.023, doi:10.1016/j.solmat.2010.09.023. generators under non-STC conditions, E3S Web Conf. 297 (2021), 01034, https://
[33] R. Khezzar, M. Zereg, A. Khezzar, Modeling improvement of the four parameter doi.org/10.1051/e3sconf/202129701034.
model for photovoltaic modules, Sol. Energy 110 (14) (2014) 452–462, https:// [48] O. Ayadi, R. Shadid, A. Bani-Abdullah, M. Alrbai, M. Abu-Mualla, N. Balah,
doi.org/10.1016/j.solener.2014.09.039. Experimental comparison between Monocrystalline, Polycrystalline, and Thin-film
[34] L. Shu-xian, W. Shuo, Y. Gui-hong, G. Ting-ting, A new explicit double-diode solar systems under sunny climatic conditions, Energy Rep. 8 (2022) 218–230,
modeling method based on Lambert W-function for photovoltaic arrays, Sol. https://doi.org/10.1016/j.egyr.2022.06.121.
Energy 116 (2015) 69–82, https://doi.org/10.1016/j.solener.2015.03.043. [49] G. Makrides, B. Zinsser, A. Phinikarides, M. Schubert, G.E. Georghiou, Temperature
[35] K. Chennoufi, M. Ferfra, M. Mokhlis, An accurate modelling of PV Modules based and thermal annealing effects on different photovoltaic technologies, Renew.
on two-diode model, Renew. Energy (2020). S0960-1481(20)31834-6. Energy 43 (2012) 407–417, https://doi.org/10.1016/j.renene.2011.11.046.
[36] M.G. Villalva, J.R. Gazoli, E.R. Filho, Comprehensive approach to modeling and [50] A.A. Elbaset, H. Ali, M. Abd-El Sattar, Novel seven-parameter model for
simulation of photovoltaic arrays, IEEE Trans. Power Electron. 24 (NO. 5) (2009), photovoltaic modules, Sol. Energy Mater. Sol. Cell. 130 (2014) 442–455, https://
https://doi.org/10.1109/TPEL.2009.2013862. doi.org/10.1016/j.solmat.2014.07.016.
[37] M.E.N.G. Zhuo, Z.H.A.O. Yiman, T.A.N.G. Shiqing, S.U.N. Yize, “An efficient [51] N. Enebish, D. Agchbayar, S. Dorjkhand, D. Baatar, I. Ylemj, Numerical analysis of
datasheet-based parameters extraction method for two-diode photovoltaic cell and solar cell current-voltage characteristics, Sol. Energy Mater. Sol. Cell. 29 (3) (1993)
cells model”, Renew. Energy (2020) https://doi.org/10.1016/j. 201–208, https://doi.org/10.1016/0927-0248(93)90035-2.
renene.2020.11.085.

15