1. Introduction
Solar generating systems (SGSs) are promising tools for converting renewable solar energy into electrical energy. SGSs are distinct from other tools since they have shorter setup time, simpler design, longer life, no moving mechanical parts, and no noise [
1]. SGSs can be employed either standalone with storage energy or grid-connected. The individual SGS cells generate very low power and voltage (0.5~0.65 V) [
2], so groups of standard numbers of SGS cells are connected in series to create a module, which for the same reason is connected in series and parallel groups to create an array [
3]. Bifacial SGS can yield energy from both the front side and back side, which can increase the yield energy up to 30% in comparison with traditional monofacial SGS; thus, the energy cost will decrease. Consequently, the SGS industries have recently paid attention to this technology [
4].
Three main points (TMP) on the I/V relationship of SGS are commonly declared in datasheets at regular atmospheric condition (RAC) (ambient temperature T
am = 20 °C and irradiance G = 1 kW/m
2), namely no-load voltage (V
n-l), short-circuit current (I
s-c), and voltage (V
m-p) with current (I
m-p) at maximum power (P
max). The aforesaid stated points do not suffice to characterize SGS as proven in the results of this article. The atmospheric conditions vary continuously, so a perfect I/V relationship of SGS at entire statuses is essential to assure an acceptable functioning analysis of SGS during different statuses. The I/V plots of SGSs can be modeled by one-, two-, and three-diode models (3-DM), which have five, seven, and nine unbeknown parameters (UPs), respectively. The higher the number of diodes in the SGS model, the more precise modeling of the losses [
5], so we chose 3-DM for modeling the SGS.
Meanwhile, the application of one of the newest metaheuristic algorithms for parametrizing the 3-DM of SGSs is treated in this article; similar efforts exerted in this regard are surveyed. Many studies have been directed to identify the UPs of the SGS model via numerous optimization algorithms because of its effectiveness in defining the optimum parameters to symmetrical and asymmetrical problems, such as the particle swarm approach [
1,
6,
7,
8,
9,
10], differential evolution (DE) [
11,
12,
13,
14], farmlands fertility approach [
15], support vector machine [
16], and counteraction-based sine–cosine approach [
17]. Additional modern meta-heuristic-based optimizers, e.g., cuckoo seeking optimizer [
18,
19], backtracking seeking algorithm [
20], imperialist competitive optimizer [
21], highest likelihood estimator [
22], approximating and correcting procedure [
23], Jaya approach [
24], interior seeking approach [
25], Levenberg–Marquardt optimizer [
26], coot optimizer [
27], gaining–sharing knowledge algorithm [
28], drone squadron algorithm [
29], gorilla optimizer [
30], and gray wolf optimizer [
31] have been utilized to parametrize the SGS model. All of the abovementioned articles treated parametrization of one- and/or two-diode models of SGSs, while some algorithms were employed to parametrize 3-DM of SGSs, namely the sunflower approach (SFA) [
5], manta-rays foraging approach (MRFA) [
32], Harris hawks algorithm [
33], grasshopper optimizer (GO) [
34], coyote approach (CA) [
35], whale optimizer (WO) [
36], marine predator approach [
37], and artificial electrical field optimizer [
38].
Regarding the above abridged review and complying with the no-free-launch theorem [
39], it remains a likelihood to ameliorate the parametrization of the SGS model. For this purpose, this research treats the employment of the barnacles mating optimization algorithm (BMOA) to parametrize the 3-DM of the SGSs. BMOA is one alternative of a metaheuristic; other potential alternatives include the adapted tabu search [
40], simulated annealing [
41], genetic algorithms [
42], and ant colony optimization [
43]. BMOA is chosen as it has been innovated recently [
44]; in addition, its issued results are encouraging and attest its vantage over other approaches. BMOA is inspired by the mating manner of barnacles in nature to solve optimization topics. Employment of BMOA was successful in optimum reactive power dispatch [
44] in addition to economical and emissions dispatch [
45].
The contributions of this research are: (i) Innovative employment of BMOA to optimally parametrize the 3-DM of the SGSs; (ii) comparison of BMOA with other approaches via the results of parametrizing 3-DM of the SGS commonly used in the literature; (iii) experimental validation of an actual case study of a commercial SGS module, viz. P-120; (iv) comparison of the 3-DM of the SGS with one- and two-diode models (1- and 2-DMs) via the results of parametrizing them.
The paper is organized as follows:
Section 2 displays 3-DM of the SGSs. The fitness function (FF) as well as the constraints are expressed in
Section 3. BMOA procedures are stated in
Section 4. Materials and methods are presented in
Section 5. The yielded results are discussed in
Section 6. Validations at changed conditions are presented in
Section 7. Conclusions are briefed in
Section 8.
2. Modeling the SGS via 3-DM
The nonlinearity of the 3-DM of the SGSs, which is the cause of orientation to optimization for identifying its UPs, is illustrated in this section. The 3-DM of the SGSs, as displayed in
Figure 1, is comprised of an electric current source in parallel with three diodes and shunt resistance (
). There exist series resistances (
) with the mentioned elements. The power losses inside the cell are embodied via
and
. Applying the current law of Kirchhoff on the equivalent circuit of 3-DM of the SGS leads to:
where:
is generated current of SGS cell;
is photo-produced current;
is the current passing in a diode;
is generated voltage of SGS cell;
is diode quality coefficient, which interprets various mechanisms responsible for moving carriers across the diode junction, if the transport process is wholly diffusion, and if it is mainly recombination in the depletion zone;
is diode opposite saturated current;
is thermic voltage;
is Boltzmann constant J/K;
TK is the cell temperature (K);
is the electron charge .
The module owns
series linked cells with gross module voltage
of
and the module current
equals
, the relationship in module is obtained via modifying (2) as stated in (4).
Figure 1.
Equivalent circuit of 3-DM of the SGS.
Figure 1.
Equivalent circuit of 3-DM of the SGS.
The shunt resistance current for the module
is calculated using (5).
It is recognized that
and
do not count on neither G nor T variants [
46]. Nevertheless, other variables, viz.
,
and
count on either G and/or T as detailed below.
principally counts on G and T [
47], as stated in (6), knowing that the regular cell temperature throughout testing is 25 °C.
where:
is the photo-produced current at RAC;
is the temperature factor of the current;
is the cell temperature (°C);
counts on
[
47,
48,
49], as depicted in (7).
where:
is diode reverse-saturation current at RAC;
is band-gap energy;
counts on T that can be modified via (8). The open-circuit voltage is reliant on T as stated in (9).
where:
is band-gap energy at RAC;
is no-load voltage at RAC;
is the temperature factor of voltage;
Shunt resistance counts on
[
46,
50] as stated in (10).
where
is shunt resistance at RAC.
The cell working temperature is nearly defined [
51] as depicted in (11).
where
is regular functioning cell temperature (
).
Regarding (1) to (6), it is obvious that there exist nine Ups, viz. , which need to be identified since they are not prescribed in the datasheets. BMOA is employed to optimize the Ups for owning the fittest values inside their higher and lower limits. Optimization is performed for assurance of precise modeling of SGS aimed at simulation and analysis.
3. Formulating the FF and Its Constraints
The FF to be minimized by BMOA is formulated in this section. FF works toward minimizing the root of the mean-squared errors (RMSEs) amongst the analogous measured and computed electric currents of SGS as depicted in (12).
where:
is the number of measurements;
are the computed and measured current of SGS, respectively.
While scrutinizing (1) and (2), we notice the nonlinearity in (1) as well as the existence of
on both sides of (1); this necessitates computing
numerically. The Newton–Raphson method is chosen to be utilized via rewriting (1) as zero formula
and calculating its derivative
with respect to
, then
can be computed as below:
where
is the
iteration.
The upper and lower limits of identify the constraints to which the FF is submitted.
5. Materials and Methods
Two modules are under study to assess the efficacy of the proposed BMOA-based procedure in parametrizing the 3-DM of the SGSs. The studied modules are Kyocera KC200GT and Copex P-120. The data of the first module were obtained via ETAP library [
56] and the second is experimentally employed and its specifications can be obtained from [
57]. The electrical specifications of two SGSs are revealed in
Table 1. The limits of the nine UPs of the 3-DM
are listed in
Table 2. The simulations are implemented using MATLAB installed on Fujitsu Laptop with an Intel(R) Core ™ i7-4702MQ CPU@ 2.2 GHz 2.2 GHz and RAM of 8 GB.
Figure 3 exhibits the experimental setup of P-120 SGS, which was held in September 2021 above the building of laboratories in Taif University. Firstly, in this Section, P-120 module is implemented at RAC where G and
are measured via pyranometer and thermometer, respectively. Secondly, P-120 module is implemented at various irradiances and temperatures to exhibit BMOA functioning at various conditions and the results are recorded in
Section 6. Potentiometers are employed to load this SGS progressively up to 8 A. Digital multimeters are employed to measure
and
. Twenty-one points of measurements of V/I are recorded. The equipment class and errors are revealed in
Table 3.
The ite
max is 1000.
Table 4 reveals BMOA factors (po, pl) which are defined through trials and errors as followed in all metaheuristic-based optimizers.
6. Results and Discussion
After employing BMOA, the electric currents are computed from the 3-DM; thereafter, RMSEs are calculated via (12), resulting in the smallest value of 9.22045 × 10
−2 using all measured points (AMP), which is less than that resulted from other optimizers in the literature by 35.05% to 87.73%, and when TMP is used, the resultant RMSEs are 3.55604 × 10
−18, which are less than that resulted from other optimizers in the literature by 99.91% to 99.99%, as written in
Table 5. The benefit of utilizing BMOA for identifying the nine UPs of the 3-DM of SGSs is obvious since the minimum decrease in RMSEs is 35.05% using AMP and 99.91% using TMP. The fittest values of the UPs of the 3-DM are acquired after numerous independent runs of BMOA producing minimum FF because of the stochastic nature of these optimizers. The convergence tendency of the RMSEs diagram is displayed in
Figure 4.
There are no results in the literature about the P-120 module that was experienced for our article. For validating the BMOA results, two additional optimizers are used, viz. the equilibrium optimizer (EO) [
58] and differential search optimizer (DSO) [
59] with
and
as adjusted for BMOA to associate the comparison with fairness.
Comparisons between BMOA, EO, and DSO dependent on the results show that the obtained RMSEs by BMOA own the lowest value, namely 9.22045 × 10
−2 using AMP, which is less than that which resulted from other optimizers in the literature by 11.18% to 20.73%, and when TMP is used, the resultant RMSEs are 3.55604 × 10
−18, which is less than that which resulted from other optimizers in the literature by 99.99%. The advantage of utilizing BMOA for identifying the nine UPs of the 3-DM of SGSs is clear, since the minimum decrease in RMSEs is 11.18% using AMP and 99.99% using TMP, as revealed in
Figure 5 and
Table 6.
Extracted UPs per cell of the 3-DM of the SGSs using AMP are written in
Table 7. With reference to (5) and (6), it is clear that the total module parameters, namely shunt, series resistors, and diode quality coefficient, can be computed as R
sh/module = N
ce × (R
sh/cell), R
s/module = N
ce × (R
s/cell), and M
1-3/module = N
ce × (M
1-3/cell), respectively. The RMSEs convergences are revealed in
Figure 4 and
Figure 5. The convergence variability among the different SGSs is due to different ratings of power, voltage, and current and accordingly different measurements. Additionally, the stochastic nature of the optimizer leads to convergence variability among the different SGSs.
The yielded results for the I/V plot of SGS at RAC that are extracted by BMOA using AMP and the measured values are displayed in
Figure 6a and
Figure 7a, for KC200GT and P-120 modules, correspondingly. Closeness amongst the experienced electric currents and calculated electric currents by BMOA affirms the precision of the estimated UPs of the 3-DM of the SGSs. While similar results are extracted by BMOA using TMP (
Figure 6b and
Figure 7b), there is less closeness despite the RMSEs being smaller, since RMSEs are computed only at TMP, which prove that TMP does not suffice to characterize SGS, as mentioned in
Section 1.
In the same manner, the P/V plots are displayed in
Figure 8a and
Figure 9a for KC200GT and P-120 modules using AMP, correspondingly, while
Figure 8b and
Figure 9b display the P/V plots using TMP. Once more, well-coinciding amongst 3-DM along with the analogous measured values is recognized when using AMP more than that when using TMP. Operational measures are prepared through parametric tests to attest the dynamism of the BMOA results using AMP.
Table 8 abridges the BMOA procedures along 100 autonomous runs and associated pointers, namely best, mean, worst, and standard deviation (SD) of RMSE values. It can be said that the lesser SDs affirm the dynamism of the cropped results and manifest the delicate choice of adjusted BMOA factors.
After comparing BMOA with other optimizers to identify the nine UPs of the 3-DM of SGSs, the 3-DM of SGSs needs to be proven to have the most precise modeling of the losses by comparing it with the 1- and 2-DMs of KC200GT, as written in
Table 9. Comparisons between the 1-, 2-, and 3-DMs of SGSs dependent on the results show that the obtained RMSEs using BMOA via the 3-DM have the lowest value, namely 1.441 × 10
−2 which is less than that which resulted from the 1- and 2-DMs by 85.93% and 88.46%, respectively.