Dynamic Lévy–Brownian marine predator algorithm for photovoltaic model parameters optimization Authors: Yassine Bouteraa, Mohammad Khishe Published: 2024 ABSTRACT The dynamic and multimodal nature of photovoltaic (PV) systems makes it challenging to examine all solar photovoltaic characteristics. Consequently, this study recommends a recently developed optimization method called the marine predator algorithm (MPA) for developing reliable PV models. In the traditional MPA, the two main search processes are Lévy flight (LF) and Brownian walk (BW), and the switch across them is unpredictable. This is while the transition between these two mechanisms is naturally continuous and dynamic. To rectify the limitation mentioned above, this research paper presents an innovative, dynamic shift function that effectively modulates the interplay that exists between the BW and LF procedures. By enhancing the changeover pattern between the primary phases of MPA, the suggested dynamic walk substantially boosts the performance of MPA. The dynamic Lévy-Brownian MPA (DLBMPA) is also made to be resilient in dealing with the parameterization limitations of PV Modeling approaches by using a constraint handling technique. The performance of DLBMPA is tested using ten popular optimization methods. Employing the DLBMPA achieved an average RMSE of 9.7 × 10− 4 in the parameter estimation across a number of multiple PV models, including the SDM, DDM, and TDM, where out of the ten optimization algorithms experimented, this was statistically significant (p < 0.05) better. In terms of averaged computation time, DLBMPA was 13 ms and still showed high accuracy in dealing with different irradiance and temperature levels. These improvements allow for MBPA to be credited as having a high efficiency when estimating the PV parameters since its speed of convergence and accuracy level surpass the previous techniques used. Introduction Because it has many applications, solar energy differs from other alternative resources. PV systems can utilize solar resources to generate electrical energy to satisfy the demand for electric power1. Solar heat is used in a variety of applications, including direct use of captured solar energy in heating rooms, advanced heating systems, carbon footprint minimization systems, and many industrial applications2. Obtaining accurate data remains challenging due to the use of outdated and non-existent equipment. The performance of PV models can be enhanced by trying them out in different locations and also by understanding the conditions that affect them3. The present study analyzes some parameters of the module models, such as the SDM, DDM, TDM, and PV module models, among others. The ongoing research will address the features inherent in each of the models and their related individual merits and demerits. It is critical to analyze in-depth what features of the above class of systems are relevant in relation to PV systems in particular4. Researchers have proposed a number of meta-optimization techniques for solar cell or module parameter estimation5. However, the objective function computation requires precise data from photovoltaic cells, which manufacturers do not supply. The PV datasheet was utilized in several articles to find or create a unique objective function and employ metaheuristic methods to obtain PV characteristics. Only a few examples of optimization are DE6, MSFL7, and OBF8. In this research, MPA9 is used to determine the best value for mentioned parameters. The two primary reasons for selecting MPA are as follows: The theorem of No Free Lunch10 states that it is impossible to find a single method that can universally solve all optimization problems. In this paper, the MPA as an approach to optimization problems, especially in the evaluation of parameters of PV models, is examined, and more positively, the view that all methods have places where they are most useful is upheld.The MPA’s merit: The MPA has outperformed over 20 other known optimization methods in a variety of assessment contexts, including designed test functions, high-dimensional engineering problems, practical problems, IEEE CEC 2017 competition benchmarks, etc. It stands out, especially in faster convergence, better toleration, and the ability to break through local optimum stagnation11.The use of optimization techniques for the estimation of PV model parameters presents two significant difficulties: typologizing multiple global/local optima and protecting this optima up to the final period. MPA does not contain any agent to address the two issues stated above on its applicability as a general-purpose optimizer. The theorem of No Free Lunch10 states that it is impossible to find a single method that can universally solve all optimization problems. In this paper, the MPA as an approach to optimization problems, especially in the evaluation of parameters of PV models, is examined, and more positively, the view that all methods have places where they are most useful is upheld. The theorem of No Free Lunch10 states that it is impossible to find a single method that can universally solve all optimization problems. In this paper, the MPA as an approach to optimization problems, especially in the evaluation of parameters of PV models, is examined, and more positively, the view that all methods have places where they are most useful is upheld. The MPA’s merit: The MPA has outperformed over 20 other known optimization methods in a variety of assessment contexts, including designed test functions, high-dimensional engineering problems, practical problems, IEEE CEC 2017 competition benchmarks, etc. It stands out, especially in faster convergence, better toleration, and the ability to break through local optimum stagnation11. The MPA’s merit: The MPA has outperformed over 20 other known optimization methods in a variety of assessment contexts, including designed test functions, high-dimensional engineering problems, practical problems, IEEE CEC 2017 competition benchmarks, etc. It stands out, especially in faster convergence, better toleration, and the ability to break through local optimum stagnation11. The use of optimization techniques for the estimation of PV model parameters presents two significant difficulties: typologizing multiple global/local optima and protecting this optima up to the final period. MPA does not contain any agent to address the two issues stated above on its applicability as a general-purpose optimizer. The use of optimization techniques for the estimation of PV model parameters presents two significant difficulties: typologizing multiple global/local optima and protecting this optima up to the final period. MPA does not contain any agent to address the two issues stated above on its applicability as a general-purpose optimizer. In this study, the authors put forth a unique and unprecedented solution that has the potential to resolve some of the persistent issues in PV model optimization. In particular, it proposes a solution that bridges an existing gap by suggesting a dynamic shift function that strengthens the local search and global search optimally, enhancing the efficacy of the MPA technique. The approach described in this paper takes into consideration the limitations of existing methods. It emphasizes the need for a less rigid and more adaptive optimization policy in order to reduce problems such as excessive specialization in LF and BW as well as the absence of transition in the search phases. In other words, conventional algorithms possess certain limitations that the authors wish to attempt to rectify. To summarize, the following are the paper’s significant contributions: In order to improve the original MPA, a DLB concept was developed.To address the constraints of PV model parameter estimations, DLBMPA presents a new constraint-handling technique.This technique is used to evaluate parameters for the SW255, SM55, and KC200GT datasheets under varied environmental conditions.DLBMPA is compared to 10 well-known optimizers. In order to improve the original MPA, a DLB concept was developed. In order to improve the original MPA, a DLB concept was developed. To address the constraints of PV model parameter estimations, DLBMPA presents a new constraint-handling technique. To address the constraints of PV model parameter estimations, DLBMPA presents a new constraint-handling technique. This technique is used to evaluate parameters for the SW255, SM55, and KC200GT datasheets under varied environmental conditions. This technique is used to evaluate parameters for the SW255, SM55, and KC200GT datasheets under varied environmental conditions. DLBMPA is compared to 10 well-known optimizers. DLBMPA is compared to 10 well-known optimizers. Complications arise along with the number of PV system parameters to be determined. These parameters are included in SDM, DDM, and TDM photovoltaic models to test the performance of the suggested technique. The results are then evaluated using other widely utilized benchmark techniques, such as EHHO12, WHHOA13, PGJAYA14, AOAHOA15, IJAYA16, QRGNMPO17, NLBBO15, STOA18, WSOA19, GWO20, and GTLBOA21. The remaining parts of the paper proceed as follows: Background information on the literature is presented in “Literature review”. “Problems definition” demonstrates various types of solar modules and cells and some necessary mathematical formulations. In “Proposed model”, the authors present and describe the DLBMPA approach to move the bias toward the consensus. “Simulation” contains the presentation of simulation results. “Simulation” addresses, in the best way, the analysis and synthesis of research results obtained in simulation, and “Conclusion” presents conclusions. Literature review Many methods have been developed to estimate photovoltaic cell/module parameters. These methods may be split into two broad categories: heuristic and deterministic22. Deterministic techniques rely on analytical or iterative approaches. In analytical approaches, parameter estimates are created using PV datasheets or I–V characteristic curves. In addition to RSS23, Lambert-W24, and OSMP25 methodologies, there are a variety of other analytical methods. As the number of PV model parameters increases, so does the computational complexity of these approaches, which entail non-linear computing models to identify the unidentified parameters26. Non-linear equations can be simplified using a series of simplifying assumptions27. Inaccurate estimations can be made by using a fixed value for Rs and Rsh, for instance, or by omitting them entirely from the problem definition. Deterministic iterative procedures require a sequence of iterations or trials to determine parameters28. When using iterative techniques, the variables in the system should be convex, continuous, and differentiable. When employing iterative approaches, finding the proper initial numbers is critical to avoiding local optima. Recently, the focus has shifted to heuristic techniques, which require tackling an optimization challenge to reveal the PV model’s parameters29. The creation of an objective function can yield previously unknown parameters that degrade its performance. A heuristic approach views parameterization as a black-box problem that can be solved with no hypotheses about the mathematical model. Recently, different metaheuristic techniques have proved successful in deriving PV parameters. As you can see in Table 1, each method has its merits and disadvantages. Table 1Meta-heuristic methodologies for PV cell/module parameter determination: benefits and shortcomings.YearMethodsReferencesSDMDDMTDMModule/cellDrawbacksBenefits2011PSA 30 √√RTC. FranceNot expressedRobustness and flexibility2011DE 31 √RTC. FranceSensitivityMutation and exploitationExplorationPrecision various conditions2012SA 32 √√RTC. FranceSensitivityConvergence rateRobustness and implementation2012HS 33 √√RTC. FranceExplorationLocal optima stagnationImplementationEffectiveness2013BMO 34 √√RTC. FranceComplexityNumber of parametersAccuracyConvergence rate2013CS 35 √RTC. FranceConvergence rateAccuracy and exploration2013ABO 36 √√RTC. FranceConvergenceLocal optima stagnationFlexibilityBalance local/global optima2013BFA 8 √SM55Convergence rateAccuracy and implementation2014MPCOA 37 √√RTC. FranceNot expressedAccuracy and balancing2014TLBO 38 √√RTC. FranceAccuracyComplexityConvergence rateImplementation2014ABC 39 √√Not ExpressedComplexitySensitivityLocal optima stagnationAccuracyLow noise sensitivityRobustness2015FPA 40 √√KC200GTLow accuracyExploitation and convergence2015AIS 41 √SM55 PVNot expressedAccuracy and convergence2016FWA 42 √KC200GTAccuracyLocal optima stagnationConvergence rateBalancing behaviourLow complexity2017WDO 43 √√KC200GTSensitivityAccuracy and explorationConvergence rate2017IAIO 44 √Not ExpressedNot expressedConvergence and accuracy2017WOA 45 √√KC200GTExplorationLocal optima stagnationComputational complexityConvergence rateExploitation2017Jaya algorithm 16 √√RTC. FranceLocal optima stagnationLow solution qualitySimple structure and robustnessConvergence rate2017ICA 46 √√KC200GTConvergence rateLocal optima stagnationComplexityAccuracy2017ER-WCA 47 √√KC200GTMedium solution qualitySuitabilityPrecision various conditionsGlobal/local balancingConvergence rate2018MLBSA 48 √√RTC. FranceNot expressedBalancing and accuracyReliability and complexity2018ISCE 49 √√RTC. FranceCapability to changeOperating circumstancesRobustnessAccuracyExplorationConvergence rate2019GWO 50 √KC200GTConvergence rateRobustness and accuracyFlexibility2019SSA 51 √TITAN-12-50Not expressedStabilityconvergenceEscape local optimum2019LCOA 52 √√RTC. FranceStability and robustnessLocal optima stagnationAccuracyConvergence rate2021IEOA 53 √√√RTC. FranceNot expressedDeeper exploitationReliability and precision2021WHHOA 13 √√√SM55KC200GTComplexity and time burdenConvergence rateReliability and accuracy2021ACSA 54 √RTC. FranceNot expressedGlobal exploration and precisionAdaptive explorationConvergence rate2022HWDFOA 55 √Real experimental dataComplexityTime burdenAccuracy and feasibilityConvergence rate2022SDOA 56 √PVM 752GaAsLack of experimentTime complexityAccuracy2022BO 57 √√RTC. FranceNot expressedRobustnessConsistency of solution Table 1Meta-heuristic methodologies for PV cell/module parameter determination: benefits and shortcomings.YearMethodsReferencesSDMDDMTDMModule/cellDrawbacksBenefits2011PSA 30 √√RTC. FranceNot expressedRobustness and flexibility2011DE 31 √RTC. FranceSensitivityMutation and exploitationExplorationPrecision various conditions2012SA 32 √√RTC. FranceSensitivityConvergence rateRobustness and implementation2012HS 33 √√RTC. FranceExplorationLocal optima stagnationImplementationEffectiveness2013BMO 34 √√RTC. FranceComplexityNumber of parametersAccuracyConvergence rate2013CS 35 √RTC. FranceConvergence rateAccuracy and exploration2013ABO 36 √√RTC. FranceConvergenceLocal optima stagnationFlexibilityBalance local/global optima2013BFA 8 √SM55Convergence rateAccuracy and implementation2014MPCOA 37 √√RTC. FranceNot expressedAccuracy and balancing2014TLBO 38 √√RTC. FranceAccuracyComplexityConvergence rateImplementation2014ABC 39 √√Not ExpressedComplexitySensitivityLocal optima stagnationAccuracyLow noise sensitivityRobustness2015FPA 40 √√KC200GTLow accuracyExploitation and convergence2015AIS 41 √SM55 PVNot expressedAccuracy and convergence2016FWA 42 √KC200GTAccuracyLocal optima stagnationConvergence rateBalancing behaviourLow complexity2017WDO 43 √√KC200GTSensitivityAccuracy and explorationConvergence rate2017IAIO 44 √Not ExpressedNot expressedConvergence and accuracy2017WOA 45 √√KC200GTExplorationLocal optima stagnationComputational complexityConvergence rateExploitation2017Jaya algorithm 16 √√RTC. FranceLocal optima stagnationLow solution qualitySimple structure and robustnessConvergence rate2017ICA 46 √√KC200GTConvergence rateLocal optima stagnationComplexityAccuracy2017ER-WCA 47 √√KC200GTMedium solution qualitySuitabilityPrecision various conditionsGlobal/local balancingConvergence rate2018MLBSA 48 √√RTC. FranceNot expressedBalancing and accuracyReliability and complexity2018ISCE 49 √√RTC. FranceCapability to changeOperating circumstancesRobustnessAccuracyExplorationConvergence rate2019GWO 50 √KC200GTConvergence rateRobustness and accuracyFlexibility2019SSA 51 √TITAN-12-50Not expressedStabilityconvergenceEscape local optimum2019LCOA 52 √√RTC. FranceStability and robustnessLocal optima stagnationAccuracyConvergence rate2021IEOA 53 √√√RTC. FranceNot expressedDeeper exploitationReliability and precision2021WHHOA 13 √√√SM55KC200GTComplexity and time burdenConvergence rateReliability and accuracy2021ACSA 54 √RTC. FranceNot expressedGlobal exploration and precisionAdaptive explorationConvergence rate2022HWDFOA 55 √Real experimental dataComplexityTime burdenAccuracy and feasibilityConvergence rate2022SDOA 56 √PVM 752GaAsLack of experimentTime complexityAccuracy2022BO 57 √√RTC. FranceNot expressedRobustnessConsistency of solution Meta-heuristic methodologies for PV cell/module parameter determination: benefits and shortcomings. Sensitivity Mutation and exploitation Exploration Precision various conditions Sensitivity Convergence rate Exploration Local optima stagnation Implementation Effectiveness Complexity Number of parameters Accuracy Convergence rate Convergence Local optima stagnation Flexibility Balance local/global optima Accuracy Complexity Convergence rate Implementation Complexity Sensitivity Local optima stagnation Accuracy Low noise sensitivity Robustness Accuracy Local optima stagnation Convergence rate Balancing behaviour Low complexity Accuracy and exploration Convergence rate Exploration Local optima stagnation Computational complexity Convergence rate Exploitation Local optima stagnation Low solution quality Simple structure and robustness Convergence rate Convergence rate Local optima stagnation Complexity Accuracy Suitability Precision various conditions Global/local balancing Convergence rate Balancing and accuracy Reliability and complexity Capability to change Operating circumstances Robustness Accuracy Exploration Convergence rate Robustness and accuracy Flexibility Stability convergence Escape local optimum Stability and robustness Local optima stagnation Accuracy Convergence rate Deeper exploitation Reliability and precision SM55 KC200GT Convergence rate Reliability and accuracy Global exploration and precision Adaptive exploration Convergence rate Complexity Time burden Accuracy and feasibility Convergence rate Lack of experiment Time complexity Robustness Consistency of solution They are well founded on the fact that the use of metaheuristics is more efficient and accurate than the use of deterministic methods. However, it is important to highlight specific approaches that require a large number of iterations so that specific results can be obtained over and over again through repetitive usage. As such, this study has introduced the DLBMPA method and amply illustrated what has been done through comparative studies with ten popular optimizers together, placing a significant input in the area of PV modeling as it presents a new way of improving the precision and efficiency of the assessment of parameters in different operational conditions. For this study, the DLBMPA technique was tested on three different industrial panels: SW255, KC200GT, and SM55. This study noted the changes in irradiance and temperature during the assessment of the performance. Other factors were held constant, except that the temperature and irradiance changed the diode and saturation current. The results obtained have shown remarkable performance by the algorithm and fast speed in convergence. Problems definition The majority of literature on PV cells and modules refers to them as diode models. The I–V family curves of diodes and photovoltaic cells possess similar characteristics, as these are microelectronic devices having an exponential region. Each diode-based model contains a number of parameters that have to be estimated, which is not the known set of parameters. Since the parameters of the photovoltaic modules change with time, as well as the amount of solar radiation received by the modules, controllers have to keep in mind the prediction of unknown parameters. Single diode model (SDM) SDMs are often favored as they are not so tedious to use. A standard SDM circuit is presented in Fig. 1. The model comprises several components, with the series and shunt resistance loss components incorporated in the current feed offsetting the ohmic running losses and the effects of the semiconductor p-n junctions. Fig. 1A typical SDM circuit model. A typical SDM circuit model. The optimal diode configuration of the SDM is produced by the coupling of diode currents with a non-physical characteristic. The Iout can be calculated by Eq. (1)16:1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} & {I_{out}}={I_{ph}} - {I_{shunt}} - {I_d}={I_{ph}} - \frac{{{I_{out}}{R_S}+{V_{out}}}}{{{R_{shunt}}}}+{I_{sat}}\left[ {1 - {e^{q(\frac{{{I_{out}}{R_S}+{V_{out}}}}{{nkT}})}}} \right] \\ & {I_{ph}}=({I_{p{h_{{\text{standard}}}}}}+{k_i}\Delta T) \times \frac{G}{{{G_{{\text{standard}}}}}} \\ & {I_{sat}}=\frac{{{k_i}\Delta T+{I_{sa{t_{{\text{standard}}}}}}}}{{{e^{[\frac{{({k_v}\Delta T+{V_{O{C_{{\text{standard}}}}}}) \times q}}{{N \times nkT}}] - 1}}}} \\ \end{aligned}$$\end{document} Currents are designated by their symbols: the current of shunt resistance current (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I_{shunt}}$$\end{document}), diode (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I_d}$$\end{document}), Boltzmann constant (k), reverse saturation (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I_{sat}}$$\end{document}), diode ideality (n), electric charge (q), and cell temperature (T) (Kelvin). Equation (2) can be used to express the SDM parameters that are currently unknown16.2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varphi =[{R_s}{\text{ }}{I_{sat}}{\text{ }}{R_{shunt}}{\text{ }}{I_{ph}}{\text{ }}n]$$\end{document} DDM The utilization of two diodes parallel to the current source in the DDM offers enhanced reliability in assessing the adverse effects of the p–n junction, particularly when exposed to low levels of radiation. A secondary diode is used in that it is necessary to imitate the effects coming from a charge donor and being a charge recombination-dominated region. Figure 2 graphically illustrates the DDM process reconstruction, stating the current stage described in Eq. (3)16:3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} {I_{out}} & ={I_{ph}} - {I_{shunt}} - {I_{d1}} - {I_{d2}} \\ & ={I_{ph}} - \frac{{{I_{out}}{R_S}+{V_{out}}}}{{{R_{shunt}}}}+{I_{sat1}}\left[ {1+{e^{q(\frac{{{I_{out}}{R_S}+{V_{out}}}}{{{n_1}kT}})}}} \right]+{I_{sat2}}\left[ {1+{e^{q(\frac{{{I_{out}}{R_S}+{V_{out}}}}{{{n_2}kT}})}}} \right] \\ \end{aligned}$$\end{document} The terms Isat1 and Isat2 represent the saturation and diffusion currents, respectively, with n1 and n2 representing the optimal transmission and recombination coefficients for the diode, respectively16.4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varphi =[{R_s}{\text{ }}{I_{sat1}}{\text{ }}{I_{sat2}}{\text{ }}{R_{shunt}}{\text{ }}{I_{ph}}{\text{ }}{n_1}{\text{ }}{n_2}]$$\end{document} Fig. 2A typical DDM circuit model. A typical DDM circuit model. Three-diode model The TDM model exhibits greater realism compared to previous models, such as the SDM and DDM, due to the incorporation of leakage current and granularity limitations; however, this complicates the model. Three diodes are linked in series with the current source in Fig. 3. Fig. 3A typical TDM circuit model. A typical TDM circuit model. The determination of the output current is defined by Eq. (5)16:5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} {I_{out}} & ={I_{ph}} - {I_{shunt}} - {I_{d1}} - {I_{d2}} - {I_{d3}} \\ & ={I_{ph}} - \frac{{{I_{out}}{R_S}+{V_{out}}}}{{{R_{shunt}}}}+{I_{sat1}}\left[ {1+{e^{q(\frac{{{I_{out}}{R_S}+{V_{out}}}}{{{n_1}kT}})}}} \right] \\ & \quad +{I_{sat2}}\left[ {1+{e^{q(\frac{{{I_{out}}{R_S}+{V_{out}}}}{{{n_2}kT}})}}} \right]+{I_{sat3}}\left[ {1+{e^{q(\frac{{{I_{out}}{R_S}+{V_{out}}}}{{{n_3}kT}})}}} \right] \\ \end{aligned}$$\end{document} Nine unknown parameters in the model are represented by a vector with the form given by Eq. (6)16.6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varphi =[{R_s}{\text{ }}{I_{sat1}}{\text{ }}{I_{sat2}}{\text{ }}{I_{sat3}}{\text{ }}{R_{shunt}}{\text{ }}{I_{ph}}{\text{ }}{n_1}{\text{ }}{n_2}{\text{ }}{n_3}]$$\end{document} PV module Figure 4 shows the series or parallel connection of N cells with a specific voltage and current. The panel is a common acronym for these kinds of solar cells. Equation (7) is used to calculate the PV module’s output current13:7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${I_{out}}={I_{ph}} - \frac{{{V_{out}}+{I_{out}}{R_S} \times N}}{{{R_{shunt}} \times N}} - {I_{sat}}\left[ {{e^{q(\frac{{{V_{out}}+{I_{out}}{R_S} \times N}}{{N \times nkT}})}} - 1} \right]$$\end{document} Fig. 4A typical PV circuit model. A typical PV circuit model. N denotes the total number of PV cells in the solar panel. Each model has some unknown parameters that are to be considered; some further research is needed. This equation employs a number of estimated parameters, utilizing a metric known as current relative error to analyze precision. These estimates were included in the analysis to approximate the parameters of the proposed models (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\operatorname{Re} {l_{error}}$$\end{document})13:8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\operatorname{Re} {l_{error}}=\frac{{{I_{out}} - {I_{est}}}}{{{I_{est}}}}$$\end{document} If the current relative error score is lower than the previous one, the parameter estimates will be improved more. Objective function Optimization approaches, based on metaheuristic methodology, yield the design variables of various PV models. This approach has led to the discovery of additional parameters, with the goal of reducing the objective function to address this issue. The objective function is most commonly defined as the squared difference between the projected and actual outputs. RMSE is a widely employed concept in this subject, which is employed to model the objective function13:9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Fit(\varphi )=RMSE(\varphi )=\sqrt {\frac{1}{{{N_s}}} \times \sum\limits_{{k=1}}^{{{N_s}}} {{f_{type}}{{({V_{out}},{I_{out}},\varphi )}^2}} }$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\varphi$$\end{document} represents the unknown parameter vector, type represents the PV model type, and Ns represents the quantity of measurement values. The error function, denoted by the symbol ftype, can be defined in several ways depending on the type of PV model13.10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${f_{SDM}}({V_{out}},{I_{out}},\varphi )={\varphi _1} - {\varphi _4} \times \left[ {{e^{q(\frac{{{V_{out}}+{I_{out}}{\varphi _2}}}{{{\varphi _5}kT}})}} - 1} \right] - \frac{{{V_{out}}+{I_{out}}{\varphi _2}}}{{{\varphi _3}}} - {I_{out}}$$\end{document}11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} & {f_{DDM}}({V_{out}},{I_{out}},\varphi )={\varphi _1} - {\varphi _4} \times \left[ {{e^{q(\frac{{{V_{out}}+{I_{out}}{\varphi _2}}}{{{\varphi _6}kT}})}} - 1} \right] \\ & \quad - {\varphi _5} \times \left[ {{e^{q(\frac{{{V_{out}}+{I_{out}}{\varphi _2}}}{{{\varphi _7}kT}})}} - 1} \right] - \frac{{{V_{out}}+{I_{out}}{\varphi _2}}}{{{\varphi _3}}} - {I_{out}} \\ \end{aligned}$$\end{document}12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} & {f_{TDM}}({V_{out}},{I_{out}},\varphi )={\varphi _1} - {\varphi _4} \times \left[ {{e^{q(\frac{{{V_{out}}+{I_{out}}{\varphi _2}}}{{{\varphi _7}kT}})}} - 1} \right] - {\varphi _5} \times \left[ {{e^{q(\frac{{{V_{out}}+{I_{out}}{\varphi _2}}}{{{\varphi _8}kT}})}} - 1} \right] \\ & \quad - {\varphi _6} \times \left[ {{e^{q(\frac{{{V_{out}}+{I_{out}}{\varphi _2}}}{{{\varphi _9}kT}})}} - 1} \right] - \frac{{{V_{out}}+{I_{out}}{\varphi _2}}}{{{\varphi _3}}} - {I_{out}} \\ \end{aligned}$$\end{document}13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${f_{Panel}}({V_{out}},{I_{out}},\varphi )={\varphi _1} - \left[ {{e^{(\frac{{{V_{out}}+{I_{out}}{\varphi _2} \times N}}{{{\varphi _5}kTN}}) \times q}} - 1} \right] \times {\varphi _4} - \frac{{{V_{out}}+{I_{out}}{\varphi _2} \times N}}{{{\varphi _3} \times N}} - {I_{out}}$$\end{document} Note that the following equations can express the vectors of unknown parameters13:14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varphi _{SDM}}=[{\varphi _1}{\text{ }}{\varphi _2}{\text{ }}{\varphi _3}{\text{ }}{\varphi _4}{\text{ }}{\varphi _5}]$$\end{document}15\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varphi _{DDM}}=[{\varphi _1}{\text{ }}{\varphi _2}{\text{ }}{\varphi _3}{\text{ }}{\varphi _4}{\text{ }}{\varphi _5}{\text{ }}{\varphi _6}{\text{ }}{\varphi _7}]$$\end{document}16\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varphi _{TDM}}=[{\varphi _1}{\text{ }}{\varphi _2}{\text{ }}{\varphi _3}{\text{ }}{\varphi _4}{\text{ }}{\varphi _5}{\text{ }}{\varphi _6}{\text{ }}{\varphi _7}{\text{ }}{\varphi _8}{\text{ }}{\varphi _9}]$$\end{document}17\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\varphi _{Panel}}=[{\varphi _1}{\text{ }}{\varphi _2}{\text{ }}{\varphi _3}{\text{ }}{\varphi _4}{\text{ }}{\varphi _5}]$$\end{document} Proposed model The mathematical equation for MPA is first covered in this section, followed by the introduction of the DLBMPA principle, and then DLBMPA is utilized to determine the PV cell’s unknown parameters. MPA algorithm The MPA can be classified as a swarm intelligence-based methodology, similar to a variety of metaheuristic techniques. Therefore, initial answers are often distributed around the search region for assessment using the method outlined in the following manner9:18\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${X_0}={X_{\hbox{min} }}+rand({X_{\hbox{max} }} - {X_{\hbox{min} }})$$\end{document} The variable “rand” represents a random vector ranging from 0 to 1. The variables’ upper and lower boundaries are denoted by Xmax and Xmin, respectively. In summary, the most effective MPA solution to date is recognized as a prominent predator in the development of Elite, denoted in Eq. (19). Matrix arrays will be used to locate Prey based on data about their location, which must be made clear9.19\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathbf{Elite}}={\left[ \begin{gathered} X_{{1,1}}^{I}{\text{ }}X_{{1,2}}^{I}{\text{ }} \cdots {\text{ }}X_{{1,d}}^{I} \hfill \\ X_{{2,1}}^{I}{\text{ }}X_{{2,2}}^{I}{\text{ }} \cdots {\text{ }}X_{{2,d}}^{I} \hfill \\ {\text{ }} \vdots {\text{ }} \vdots {\text{ }} \vdots {\text{ }} \vdots \hfill \\ X_{{n,1}}^{I}{\text{ }}X_{{n,2}}^{I}{\text{ }} \cdots {\text{ }}X_{{n,d}}^{I} \hfill \\ \end{gathered} \right]_{n \times d}}$$\end{document} where the symbol XI denotes the most superior predator, which is iterated n times, which represents the overall count of search agents, whereas the variable d denotes the total count of dimensions, the composition of the Elite population is modified when the superior predator is replaced by a more proficient searching agent. Prey, akin to Elite, can be conceptualized as a dimensional matrix. Agents utilize a grid to monitor their whereabouts, wherein the initial Prey is established via initialization. Equation (20) is employed to denote the Prey in the following manner9:20\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathbf{Prey}}={\left[ \begin{gathered} X_{{1,1}}^{{}}{\text{ }}X_{{1,2}}^{{}}{\text{ }} \cdots {\text{ }}X_{{1,d}}^{{}} \hfill \\ X_{{2,1}}^{{}}{\text{ }}X_{{2,2}}^{{}}{\text{ }} \cdots {\text{ }}X_{{2,d}}^{{}} \hfill \\ {\text{ }} \vdots {\text{ }} \vdots {\text{ }} \vdots {\text{ }} \vdots \hfill \\ X_{{n,1}}^{{}}{\text{ }}X_{{n,2}}^{{}}{\text{ }} \cdots {\text{ }}X_{{n,d}}^{{}} \hfill \\ \end{gathered} \right]_{n \times d}}$$\end{document} The value of X i, j denotes the jth dimension of the ith predator. The optimization procedure has a strong and accurate correlation with the Elite and Prey matrices. The MPA consists of three distinct phases, namely the Low-Velocity Ratio (LVR), High-Velocity Ratio (HVR), and Unit-Velocity Ratio (UVR) phases. The three stages mentioned above have been explained in detail in9. Phase 1 HVR: 21\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} While{\text{ Iteration < max(Iteration)}} \times \frac{1}{3} \hfill \\ {\mathbf{stepsiz}}{{\mathbf{e}}_{\mathbf{i}}}={{\mathbf{R}}_{{\mathbf{BM}}}} \otimes ({\mathbf{Elit}}{{\mathbf{e}}_{\mathbf{i}}} - {\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}} \otimes {{\mathbf{R}}_{{\mathbf{BM}}}}){\text{ }} \hfill \\ {\text{ }}{\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}={\text{ }}{\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}+{\mathbf{stepsiz}}{{\mathbf{e}}_{\mathbf{i}}} \otimes P.{\mathbf{R}} \hfill \\ {\text{i = 1, }}…{\text{, n}} \hfill \\ \end{gathered}$$\end{document} Phase 1 HVR: Phase 1 HVR: Phase 2 UVR: 22\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\text{while }}\frac{{Max(Iteration)}}{3}{\text{ < Iteration < }}\frac{{2Max(Iteration)}}{3} \hfill \\ {\text{For: the first }} \times \frac{{{\text{searching agents}}}}{{\text{2}}} \hfill \\ {\mathbf{stepsiz}}{{\mathbf{e}}_{\mathbf{i}}}={{\mathbf{R}}_{{\mathbf{LF}}}} \otimes ({\mathbf{Elit}}{{\mathbf{e}}_{\mathbf{i}}} - {\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}} \otimes {{\mathbf{R}}_{{\mathbf{LF}}}}){\text{ }} \hfill \\ {\text{ }}{\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}={\text{ }}P.{\mathbf{R}} \otimes {\mathbf{stepsiz}}{{\mathbf{e}}_{\mathbf{i}}}+{\mathbf{Pre}}{{\mathbf{y}}_{{\mathbf{i}} - 1}} \hfill \\ {\text{ i = 1, }}…{\text{, n/2}} \hfill \\ \end{gathered}$$\end{document} 23\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\mathbf{stepsiz}}{{\mathbf{e}}_{\mathbf{i}}}={{\mathbf{R}}_{{\mathbf{BM}}}} \otimes ({\mathbf{Elit}}{{\mathbf{e}}_{\mathbf{i}}} \otimes {{\mathbf{R}}_{{\mathbf{BM}}}} - {\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}){\text{ }} \hfill \\ {\text{ }}{\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}={\text{ }}P.{\text{CF}} \otimes {\mathbf{stepsiz}}{{\mathbf{e}}_{\mathbf{i}}}+{\mathbf{Elit}}{{\mathbf{e}}_{\mathbf{i}}} \hfill \\ {\text{CF=(1-}}\frac{{Iteration}}{{Maximum(Iteration)}}{{\text{)}}^{2 \times \frac{{Iteration}}{{Maximum(Iteration)}}}} \hfill \\ {\text{i = n/2, }}…{\text{, n}} \hfill \\ \end{gathered}$$\end{document} Phase 2 UVR: Phase 2 UVR: Phase 3 LVR: 24\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\text{While Iteration > }}0.66{\text{ Max (Iteration)}} \hfill \\ {\mathbf{stepsiz}}{{\mathbf{e}}_{\mathbf{i}}}={{\mathbf{R}}_{{\mathbf{LF}}}} \otimes ({{\mathbf{R}}_{{\mathbf{LF}}}} \otimes - {\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}+{\mathbf{Elit}}{{\mathbf{e}}_{\mathbf{i}}}){\text{ i = 1, }}…{\text{, n}} \hfill \\ {\text{ }}{\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}={\text{ }}P.{\text{CF}} \otimes {\mathbf{stepsiz}}{{\mathbf{e}}_{\mathbf{i}}}+{\mathbf{Elit}}{{\mathbf{e}}_{\mathbf{i}}} \hfill \\ {\text{CF=(1-}}\frac{{Iteration}}{{Max(Iteration)}}{{\text{)}}^{\frac{{2Iteration}}{{Max(Iteration)}}}} \hfill \\ \end{gathered}$$\end{document} Phase 3 LVR: Phase 3 LVR: To complicate matters further, local optimum avoidance operators [84], such as Fish Aggregating Devices (FADs) or eddy formation influences, might alter the behavior of marine predators. In order to prevent the simulation from becoming stuck in a local optimum, lengthier hops should be taken. The following is a summary of the impact of FAD:25\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{ }}{\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}=\left\{ \begin{gathered} [(r - 1) \times FADs+r]({\mathbf{Pre}}{{\mathbf{y}}_{r2}} - {\mathbf{Pre}}{{\mathbf{y}}_{r1}})+{\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}{\text{ if r}}>FADs \hfill \\ {\text{CF}} \times {\text{[}}({{\mathbf{X}}_{\operatorname{m} ax}} - {{\mathbf{X}}_{\hbox{min} }}) \otimes {\mathbf{R}}+{{\mathbf{X}}_{\hbox{min} }}] \otimes {\mathbf{U}}+{\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}{\text{ if r}} \leqslant FADs \hfill \\ \end{gathered} \right.$$\end{document} Where the likelihood of the FAD’s effect on the optimization procedure is 0.2, the binary vector is denoted by U. After producing a random vector, the array is altered to contain a value of one if the value is greater than 0.2 and zero otherwise. Also, r is an integer between 0 and 1. Each dimension has a minimum and maximum value, denoted by Xmin and Xmax, respectively. For this scenario, random indices r1 and r2 will be employed to get entries from the Prey matrix. Marine predators can quickly and accurately recall the locations of previous successful hunts. For comparison purposes, assign more importance to the search agents of this current iteration than to those of the previous iteration. The DFSMPA The MPA is developed to describe the development of predator-prey relationships through three levels that actually exist in nature. These levels of interaction are factorized according to numerous admissible sizes of the steps that the predator would undertake so as to capture its Prey. Using the MPA, however, predicts an almost constant percentage of LFs and BMs for a predator’s life history. In the first stage, the predator is completely still; in the second stage, it operates with the BM movement; and finally, in the third stage, with the LF movement. Stage I clearly shows the Prey going in BM, while stage II clearly shows the Prey following LF. The algorithm adds that allocating a 3rd of iterations to every stage empirically improves the policy and provides marginally better results than shifting between stages or periodically repeating the stage. Considering the novelty of MPA, it might be improved by specifying how and when to perform the method for every updating phase. A seamless movement from the investigation phase to the exploitation phase is not always guaranteed by the canonical MPA. In order to solve the addressed problem, the DLB algorithm is introduced. Furthermore, the DLB can guarantee improved search trends for more effective global searching. Instead of using three discrete stages to represent the actions of Prey and predator in DSMAP-DFSMPA, it uses a fluid metaphor for gradual transitions between these stages. Continuous models enable the transfer of search agents between exploration and extraction. The value of the coefficient category (i.e., LF or BM) and Prey and Elite locations determine the step size, as stated in Eqs. (21) to (24). Applying the RBM matrix by Prey yields the step size for this stage, as shown in the first half of Eq. (21). In order to put it another way, in Eq. (22) (i.e., \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\mathbf{R}}_{{\mathbf{LF}}}} \otimes ({\mathbf{Elit}}{{\mathbf{e}}_{\mathbf{i}}} - {{\mathbf{R}}_{{\mathbf{LF}}}} \otimes {\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}})$$\end{document}), the RLF array is multiplicated by the Prey location, where the LF distribution is utilized to produce random numbers for every position. This means that four equations from Eqs. (21) to (24) are summarized in Eq. (26) as follows:26\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{gathered} {\mathbf{stepsiz}}{{\mathbf{e}}_{\mathbf{i}}}={{\mathbf{R}}_{LB2}} \otimes ({{\mathbf{R}}_{LB2}} \otimes {\mathbf{Elit}}{{\mathbf{e}}_{\mathbf{i}}} - {{\mathbf{R}}_{LB1}} \otimes {\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}){\text{ i = n/2, }}…{\text{, n}} \hfill \\ {\text{ }}{\mathbf{Pre}}{{\mathbf{y}}_{\mathbf{i}}}={\text{ }}P.{\text{CF}} \otimes {\mathbf{stepsiz}}{{\mathbf{e}}_{\mathbf{i}}}+{\mathbf{Elit}}{{\mathbf{e}}_{\mathbf{i}}} \hfill \\ {\text{CF=(1-}}\frac{{Iteration}}{{Max(Iteration)}}{{\text{)}}^{ \times \frac{{2Iteration}}{{Max(Iteration)}}}} \hfill \\ \end{gathered}$$\end{document} Equations (27) and (28), respectively, can express RLB1 and RLB2 in decreasing (W1) and increasing (W2) slopes, as well as random behavior.27\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\mathbf{R}}_{LB1}}=2.12+\frac{{rand}}{{10}}(size({{\mathbf{R}}_{LB1}}_{1})) - (\frac{{{2^{1/4}}}}{{{\text{maximum}}{{(iteration)}^{1/3}}}})$$\end{document}28\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\mathbf{R}}_{LB2}}=\frac{{rand}}{{10}}(size({{\mathbf{R}}_{LB2}}))+(0.1+\frac{{{{1.5}^{1/4}}}}{{{\text{maximum}}{{(iteration)}^{1/3}}}})$$\end{document} The shape of the LF is determined only by the parameter, resulting in distinct geometries for RLB1 and RLB2. Using a lesser value for \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha$$\end{document} results in a thorough analysis of LF or BM, displaying more prominent movements. On the other hand, choosing greater values reduces the search space and constricts the range of exploration. To develop a balanced strategy, these coefficients are used with their specific form. This entails a seamless transition between the two strategies as well as a robust pursuit parameter for considerable exploration during the early stages and an exact pursuit parameter for more precise exploitation during the latter stages. RLB2 has been chosen as the Elite coefficient, as shown in Eq. (26). As shown by the rising peak value of RLB2, the “Motion” in Fig. 5 is characterized by initially large step sizes that gradually shrink with each iteration. Due to the progressive reduction in step sizes, this pattern causes the Elite’s position to remain relatively stable in the last iterations. In order to keep the Prey from becoming stuck in local minima, DLBMPA uses a technique that involves lowering amplitude and raising step size for RLB1 across iterations. By using this strategy, the Prey can explore a wider variety of search space regions and avoid becoming caught in undesirable local optima. The following is a representation of DLBMPA in pseudo-code. Also, Fig. 5 presents the general block diagram of the DLBMPA algorithm. Fig. 5Block diagram of the DLBMPA. Block diagram of the DLBMPA. Multiple studies9,58 show limits to the range of parameters in various realistic applications. On the basis of the restriction \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$[{x_{\hbox{min} }},{x_{\hbox{max} }}]$$\end{document}, Eq. (29) can be utilized to restrict the movement of predators.29\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{x}_{\text{i}}^{\text{d}}=\text{m}\text{i}\text{n}(\text{x}_{{\text{m}\text{a}\text{x}}}^{\text{d}},\text{m}\text{a}\text{x}(\text{x}_{{\text{m}\text{i}\text{n}}}^{\text{d}},\text{x}_{\text{i}}^{\text{d}}))$$\end{document} An inventive strategy has been developed to restrict predators within their designated area: the goal function and optimal position of a predator and its Prey can be modified if the predator is within the defined range. During the hunt and its Prey, all remain within the defined search area for tissue retention. The predator will be present in this territory to Prey on Tubinidae and similar edible animals. Simulation First, the proposed DLBMPA’s performance must be evaluated by establishing PV module parameters. As a baseline, data from reference13 was used. The properties of the SM55 module have now been established under varied temperature and light circumstances. Based on the datasheets of the corresponding modules, the module behavior was effectively manipulated in accordance with the reference standards of T = 25 °C and G = 1000 W/m2. Results acquired with this method are compared to those obtained with other successful techniques. Table 2 provides the value of the parameter limits for the Photovoltaic module/cell configuration39. Five hundred iterations of DLBMPA and 25,000 functional evaluations are used to find a solution alongside the use of competing techniques. Each DLBMPA execution included the average and standard deviation of the optimal solutions obtained so far to guarantee statistically meaningful findings for this study. Initial values and other factors are summarized in Table 3. Table 2Parameters’ range.Param.PWP201RTC France PVPV moduleBoundsULULUL \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_s}(\Omega )$$\end{document} 200.5020 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_{shunt}}(\Omega )$$\end{document} 10000100050000 I ph (A) 20102Isat0Id1,Id2,Id3 (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document})500101000 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_1},{n_2}$$\end{document} 5012141 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_3}$$\end{document} 50152 Table 2Parameters’ range.Param.PWP201RTC France PVPV moduleBoundsULULUL \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_s}(\Omega )$$\end{document} 200.5020 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_{shunt}}(\Omega )$$\end{document} 10000100050000 I ph (A) 20102Isat0Id1,Id2,Id3 (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document})500101000 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_1},{n_2}$$\end{document} 5012141 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_3}$$\end{document} 50152 Parameters’ range. Table 3Setting parameters and initial values.AlgorithmParameterValueMPA 9P0.5FADs0.2WHHOA and EHHO 12,13β1.5 Q 5 F 6STOA 18c1, c22NLBBO 15E & I1Immigration probability mutation10.005GTLBOA 38Learners8QRGNMPO 34Monogamous100Promiscuous20Polygynous60Parthenogenetic10Polyandrous10PGJAYA and IJAYA 59r10.38r20.42GWO 20 a [0,2] Table 3Setting parameters and initial values.AlgorithmParameterValueMPA 9P0.5FADs0.2WHHOA and EHHO 12,13β1.5 Q 5 F 6STOA 18c1, c22NLBBO 15E & I1Immigration probability mutation10.005GTLBOA 38Learners8QRGNMPO 34Monogamous100Promiscuous20Polygynous60Parthenogenetic10Polyandrous10PGJAYA and IJAYA 59r10.38r20.42GWO 20 a [0,2] Setting parameters and initial values. RTC France solar cell DLBMPA was used to estimate five SDM parameters and ten optimization techniques, including EHHO12, WHHOA13, PGJAYA14, AOAHOA15, IJAYA16, QRGNMPO17, NLBBO15, STOA18, WSOA19, GWO20, and GTLBOA21. The outcomes are shown in Table 4. Table 4The results of SDM for RTC France.Algorithm \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_{shunt}}(\Omega )$$\end{document} Iph (A) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_s}(\Omega )$$\end{document} Id1(\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document}) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_1}$$\end{document} RMSEDLBMPA53.7064460.76066170.0351550.32171.4800059.7588e − 04PGJAYA 1453.7162700.76061120.0351720.32171.4800849.7592e − 04WHHO 1353.7064470.76066230.0352640.32171.4800069.7591e − 04EHHO 1253.7306070.76065210.0352530.32171.4801289.7591e − 04NLBBO 1553.8603710.76050000.0350760.32341.4815819.7607e − 04AOAHOA 1553.7162720.76051000.0352470.32291.4809819.7601e − 04IJAYA 1653.7472700.76069000.0351760.32171.4809809.7602e − 04STOA 1859.0107000.76059210.0341760.40001.5021801.3988e − 03QRGNMPO 1753.3541710.76069000.0350760.32861.4821809.7604e − 04WSOA 1952.2880710.76069000.0364800.30511.4859019.7600e − 04GTLBOA 2142.3606700.76170030.0286710.80671.5749801.8601e − 02GWO 2053.7064000.76066180.0352550.32201.4800339.7588e − 04 Table 4The results of SDM for RTC France.Algorithm \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_{shunt}}(\Omega )$$\end{document} Iph (A) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_s}(\Omega )$$\end{document} Id1(\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document}) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_1}$$\end{document} RMSEDLBMPA53.7064460.76066170.0351550.32171.4800059.7588e − 04PGJAYA 1453.7162700.76061120.0351720.32171.4800849.7592e − 04WHHO 1353.7064470.76066230.0352640.32171.4800069.7591e − 04EHHO 1253.7306070.76065210.0352530.32171.4801289.7591e − 04NLBBO 1553.8603710.76050000.0350760.32341.4815819.7607e − 04AOAHOA 1553.7162720.76051000.0352470.32291.4809819.7601e − 04IJAYA 1653.7472700.76069000.0351760.32171.4809809.7602e − 04STOA 1859.0107000.76059210.0341760.40001.5021801.3988e − 03QRGNMPO 1753.3541710.76069000.0350760.32861.4821809.7604e − 04WSOA 1952.2880710.76069000.0364800.30511.4859019.7600e − 04GTLBOA 2142.3606700.76170030.0286710.80671.5749801.8601e − 02GWO 2053.7064000.76066180.0352550.32201.4800339.7588e − 04 The results of SDM for RTC France. Since it is impossible to determine the correct parameter values for a method, the root-mean-squared error (RMSE) is used instead. Tables 5 and 6 present the results of the calculation of seven DDM and nine TDM parameters, respectively. These calculations were performed using the DLBMPA technique and ten benchmark techniques. Table 5Findings from the French DDM RTC.Method \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_s}(\Omega )$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_{shunt}}(\Omega )$$\end{document} Id1 (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document})Iph (A)Id2 (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document}) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_1}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_2}$$\end{document} RMSEDLBMPA0.03526555.4253200.21170.76066280.7260701.45011629.7199e − 04PGJAYA 300.03528055.7173820.20170.76071220.8838401.4400781.45119.7201e − 04WHHO 300.03522455.4253210.21780.76066340.7260711.45022729.7200e − 04EHHO 300.03526255.6383280.57170.76061000.2478601.4513001.99519.7278e − 04NLBBO 300.03527855.8018080.20340.76061010.8203381.2115781.92339.7217e − 04AOAHOA 300.03525955.8172780.21290.76050110.7204281.4410782.00009.7241e − 04IJAYA 300.03528077.7522780.00180.76062010.3834481.7110781.44109.7682e − 04STOA 300.03418043.0440080.12000.76051020.0102081.5021782.00011.6600e − 03QRGNMPO 300.03530853.4052780.31870.76060010.2142481.4621781.93639.7684e − 04WSOA 300.03550154.2901780.25510.76058030.3853081.4859011.97349.7600e − 04GTLBOA 300.02878053.3747780.00070.76168040.0000081.3740821.47083.6681e − 01GWO 300.03523355.4253250.21820.76066250.7260681.4502331.999.7225e − 04 Table 5Findings from the French DDM RTC.Method \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_s}(\Omega )$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_{shunt}}(\Omega )$$\end{document} Id1 (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document})Iph (A)Id2 (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document}) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_1}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_2}$$\end{document} RMSEDLBMPA0.03526555.4253200.21170.76066280.7260701.45011629.7199e − 04PGJAYA 300.03528055.7173820.20170.76071220.8838401.4400781.45119.7201e − 04WHHO 300.03522455.4253210.21780.76066340.7260711.45022729.7200e − 04EHHO 300.03526255.6383280.57170.76061000.2478601.4513001.99519.7278e − 04NLBBO 300.03527855.8018080.20340.76061010.8203381.2115781.92339.7217e − 04AOAHOA 300.03525955.8172780.21290.76050110.7204281.4410782.00009.7241e − 04IJAYA 300.03528077.7522780.00180.76062010.3834481.7110781.44109.7682e − 04STOA 300.03418043.0440080.12000.76051020.0102081.5021782.00011.6600e − 03QRGNMPO 300.03530853.4052780.31870.76060010.2142481.4621781.93639.7684e − 04WSOA 300.03550154.2901780.25510.76058030.3853081.4859011.97349.7600e − 04GTLBOA 300.02878053.3747780.00070.76168040.0000081.3740821.47083.6681e − 01GWO 300.03523355.4253250.21820.76066250.7260681.4502331.999.7225e − 04 Findings from the French DDM RTC. Table 6The TDM results from RTC France.Algorithm \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_s}(\Omega )$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_{shunt}}(\Omega )$$\end{document} Iph (A)Id1(\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document})Id2(\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document})Id3(\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document}) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_1}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_2}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_3}$$\end{document} RMSEDLBMPA0.03527755.4253210.760660.22890.43870.7981.4501162.002.39.7199e − 04PGJAYA140.03529155.7173710.760720.21190.80270.1001.4400931.912.49.7271e − 04WHHO130.03523555.4253210.760610.22790.43870.7991.4502382.002.29.7200e − 04EHHO120.03527455.6383170.760640.21190.57860.5721.4513921.922.19.7200e − 04NLBBO150.03519055.8018080.760580.20350.72020.0001.2105911.812.29.7220e − 04AOAHOA150.03525855.8172780.760510.21190.72030.2111.4409911.992.09.7231e − 04IJAYA160.03528177.7522780.760580.21080.78330.0001.7109891.812.29.7602e − 04STOA180.03428043.0440080.760470.22900.73010.0001.5021892.002.29.7611e − 04QRGNMPO170.03531053.4052780.760580.21870.71310.0011.4621891.922.29.7604e − 04WSOA190.03550154.2901780.760480.21520.50420.0201.4859091.972.09.7602e − 04GTLBOA210.02877053.3747780.761600.31970.36000.0001.3749891.972.11.0581e − 03GWO200.03523655.4253180.760580.22800.43880.8211.4502421.982.19.7233e − 04 Table 6The TDM results from RTC France.Algorithm \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_s}(\Omega )$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_{shunt}}(\Omega )$$\end{document} Iph (A)Id1(\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document})Id2(\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document})Id3(\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document}) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_1}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_2}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_3}$$\end{document} RMSEDLBMPA0.03527755.4253210.760660.22890.43870.7981.4501162.002.39.7199e − 04PGJAYA140.03529155.7173710.760720.21190.80270.1001.4400931.912.49.7271e − 04WHHO130.03523555.4253210.760610.22790.43870.7991.4502382.002.29.7200e − 04EHHO120.03527455.6383170.760640.21190.57860.5721.4513921.922.19.7200e − 04NLBBO150.03519055.8018080.760580.20350.72020.0001.2105911.812.29.7220e − 04AOAHOA150.03525855.8172780.760510.21190.72030.2111.4409911.992.09.7231e − 04IJAYA160.03528177.7522780.760580.21080.78330.0001.7109891.812.29.7602e − 04STOA180.03428043.0440080.760470.22900.73010.0001.5021892.002.29.7611e − 04QRGNMPO170.03531053.4052780.760580.21870.71310.0011.4621891.922.29.7604e − 04WSOA190.03550154.2901780.760480.21520.50420.0201.4859091.972.09.7602e − 04GTLBOA210.02877053.3747780.761600.31970.36000.0001.3749891.972.11.0581e − 03GWO200.03523655.4253180.760580.22800.43880.8211.4502421.982.19.7233e − 04 The TDM results from RTC France. The proposed technique has the potential to achieve fewer RMSEs in both scenarios, as demonstrated. According to the data in these tables, two further conclusions can be achieved. First, compared to the two other models, the TDM approach often obtained a lower RMSE value, making it a more realistic estimate of the performance of the RTC FRANCE cell. Furthermore, because TDM has a lower RMSE, the DLBMPA algorithm outperformed all other benchmarks when there were more unknown parameters to detect configurations. In other methods, as the number of parameters increased, usefulness diminished, and RMSE values were higher. Running the TDM algorithms 30 times ensured that the best results for each table were found to be included in the final results of the table. To provide a fair and equitable comparison, all benchmarks must commence from the same random starting point. In order to attain optimal outcomes, it is advisable to perform many repetitions of each procedure. Tables 7 and 8, and 9 exhibit the RMSE evaluated respectfully for SDM, DDM, and TDM. The Wilcoxon rank-sum test is utilized to statistically evaluate the performance enhancements obtained by DLBMPA. The selection of this non-parametric test was based on its resistance to assumptions regarding the distribution of data and its capacity to identify substantial disparities in medians. The null hypothesis posits that there is no statistically significant disparity in performance between the baseline and optimized algorithm, whereas the alternative hypothesis proposes the existence of a noteworthy enhancement. A p-value below the standard significance level of 0.05 indicates strong evidence to reject the null hypothesis, suggesting a statistically significant difference in the performance of the algorithms. The p-value quantifies the importance of this improvement, confirming the practical enhancements as well as the fact that the implemented optimization algorithm led to a statistically significant increase in performance. The N/A in these tables indicates that the Wilcoxon rank-sum test cannot compare one algorithm to itself, not that it is not available. Table 7The root-mean-squared error for SDM RTC France.MethodMinimumMeanMaximumP-valueDLBMPA9.7571e − 049.7581e − 049.7591e − 04N/APGJAYA149.7591e − 049.7597e − 049.7598e − 040.0021WHHO139.7591e − 049.7599e − 049.7611e − 040.0011EHHO129.7591e − 049.7603e − 039.7604e − 040.047NLBBO159.7597e − 049.7643e − 049.7723e − 040.001AOAHOA159.7591e − 049.7623e − 049.7633e − 040.002IJAYA169.7592e − 049.7623e − 041.0091e − 030.112STOA181.0010e − 031.2132e − 031.3433e − 030.0042QRGNMPO179.7733e − 041.1011e − 031.1214e − 030.047WSOA199.7590e − 049.7604e − 049.7610e − 040.042GTLBOA211.0581e − 031.0722e − 031.9044e − 030.0021GWO209.8591e − 049.9599e − 049.9991e − 040.0011 Table 7The root-mean-squared error for SDM RTC France.MethodMinimumMeanMaximumP-valueDLBMPA9.7571e − 049.7581e − 049.7591e − 04N/APGJAYA149.7591e − 049.7597e − 049.7598e − 040.0021WHHO139.7591e − 049.7599e − 049.7611e − 040.0011EHHO129.7591e − 049.7603e − 039.7604e − 040.047NLBBO159.7597e − 049.7643e − 049.7723e − 040.001AOAHOA159.7591e − 049.7623e − 049.7633e − 040.002IJAYA169.7592e − 049.7623e − 041.0091e − 030.112STOA181.0010e − 031.2132e − 031.3433e − 030.0042QRGNMPO179.7733e − 041.1011e − 031.1214e − 030.047WSOA199.7590e − 049.7604e − 049.7610e − 040.042GTLBOA211.0581e − 031.0722e − 031.9044e − 030.0021GWO209.8591e − 049.9599e − 049.9991e − 040.0011 The root-mean-squared error for SDM RTC France. Table 8The root-mean-squared error for DDM RTC France.MethodMinimumMeanMaximumP-valueDLBMPA9.7099e − 049.7100e − 049.7102e − 04N/APGJAYA149.7210e − 049.7326e − 049.7396e − 040.23WHHO131.0001e − 031.0100e − 031.1000e − 030.0001EHHO129.7233e − 049.7392e − 049.7411e − 040.002NLBBO159.7381e − 049.7399e − 049.7529e − 040.0001AOAHOA159.7233e − 049.7299e − 049.7386e − 040.0001IJAYA169.7270e − 049.8100e − 049.9922e − 040.47STOA181.4903e − 032.0682e − 032.3091e − 030.021QRGNMPO179.7699e − 041.0433e − 031.2871e − 030.003WSOA199.7151e − 049.8001e − 049.8411e − 040.001GTLBOA213.1233e − 024.0288e − 021.0001e − 000.001GWO201.0022e − 031.0125e − 031.1230e − 030.0001 Table 8The root-mean-squared error for DDM RTC France.MethodMinimumMeanMaximumP-valueDLBMPA9.7099e − 049.7100e − 049.7102e − 04N/APGJAYA149.7210e − 049.7326e − 049.7396e − 040.23WHHO131.0001e − 031.0100e − 031.1000e − 030.0001EHHO129.7233e − 049.7392e − 049.7411e − 040.002NLBBO159.7381e − 049.7399e − 049.7529e − 040.0001AOAHOA159.7233e − 049.7299e − 049.7386e − 040.0001IJAYA169.7270e − 049.8100e − 049.9922e − 040.47STOA181.4903e − 032.0682e − 032.3091e − 030.021QRGNMPO179.7699e − 041.0433e − 031.2871e − 030.003WSOA199.7151e − 049.8001e − 049.8411e − 040.001GTLBOA213.1233e − 024.0288e − 021.0001e − 000.001GWO201.0022e − 031.0125e − 031.1230e − 030.0001 The root-mean-squared error for DDM RTC France. Table 9The root-mean-squared error for TDM RTC France.MethodMinimumMeanMaximumP-valueDLBMPA9.6999e − 049.7002e − 049.7016e − 04N/APGJAYA149.7200e − 049.7352e − 049.7920e − 040.001WHHO131.0003e − 031.0052e − 031.0100e − 030.001EHHO129.7133e − 049.7421e − 049.7623e − 040.001NLBBO159.7233e − 042.0458e − 035.4001e − 030.012AOAHOA159.7301e − 049.7485e − 049.7611e − 040.32IJAYA169.7233e − 049.9458e − 042.1200e − 030.0001STOA189.7235e − 041.1000e − 032.9321e − 030.047QRGNMPO179.7287e − 049.9399e − 043.4527e − 030.021WSOA199.7521e − 045.5222e − 038.5979e − 030.021GTLBOA211.0529e − 031.1211e − 031.5457e − 030.0001GWO201.0022e − 031.0142e − 031.0198e − 030.012 Table 9The root-mean-squared error for TDM RTC France.MethodMinimumMeanMaximumP-valueDLBMPA9.6999e − 049.7002e − 049.7016e − 04N/APGJAYA149.7200e − 049.7352e − 049.7920e − 040.001WHHO131.0003e − 031.0052e − 031.0100e − 030.001EHHO129.7133e − 049.7421e − 049.7623e − 040.001NLBBO159.7233e − 042.0458e − 035.4001e − 030.012AOAHOA159.7301e − 049.7485e − 049.7611e − 040.32IJAYA169.7233e − 049.9458e − 042.1200e − 030.0001STOA189.7235e − 041.1000e − 032.9321e − 030.047QRGNMPO179.7287e − 049.9399e − 043.4527e − 030.021WSOA199.7521e − 045.5222e − 038.5979e − 030.021GTLBOA211.0529e − 031.1211e − 031.5457e − 030.0001GWO201.0022e − 031.0142e − 031.0198e − 030.012 The root-mean-squared error for TDM RTC France. The SDM showed that most techniques have a high level of robustness and stability. When there are more unknown factors, the DDM and TDM techniques struggle to optimize. This issue leads to the common practice in studies proposing novel techniques to exclude parameterization for the TDM. According to Tables 5, 6 and 7, DLBMPA outperformed other schemes in terms of model stability. When compared to the other algorithms, DLBMPA also fared well. Table 10 displays the relative errors and their predicted current for DLBMPA-based models. Equation (10) can be used to verify the accuracy of the parameters that have been examined. As can be seen, the projected parameters are accurate because the proportional error values are minimal. It was also demonstrated that the DLBMPA method correctly predicted the three models’ output current. Table 10The relative errors and predicted current.DataIoutVoutSDMDDMTDMRelerrorIestRelerrorIestRelerrorIest10.7518−0.2024−0.0000550.7639950.0000040.7638540.0000080.76284420.7598−0.1268−0.0007170.761515−0.0006610.761593−0.0006540.75040030.7593−0.0555−0.0000170.760212−0.0010770.760205−0.0010660.76020440.75930.00240.0003300.7600960.0003990.7600400.00039990.76004450.74780.06130.0001140.7589920.0010430.7589940.00104490.75909960.74680.11520.0011170.7579820.0010380.7579070.00102690.75809970.74480.1645−0.0000140.756994−0.0001110.756923−0.0001230.75706180.74480.21080.0022010.7550280.0009990.7551180.00087190.75611490.74230.25020.0004010.7549950.0003070.7540420.00030590.754044100.74180.28990.0003050.7525440.0002470.7526980.00026590.752690110.74930.3236−0.0011660.750289−0.0011690.750263−0.0009950.750237120.73330.3552−0.0010280.746238−0.0010340.746165−0.0008550.746006130.72530.3840−0.0010490.740095−0.0019060.738862−0.0017230.738741140.71580.41040.0007080.7263010.0009010.7261370.00130650.726991150.70320.4340−0.0005100.705822−0.0003570.7057170.00010990.705391160.66230.45670.000180.6741680.0002630.6741000.00145990.673495170.62980.47510.0016500.6307420.0016300.6307540.00355960.628628180.56080.49370.0005010.5719980.0009930.5710060.00440990.570344190.48580.5099−0.0008300.498258−0.0009040.4984360.00341010.496292200.40080.5232−0.0010410.412353−0.0012170.4124220.00573960.410505210.30330.5365−0.0011290.316192−0.0022030.3161190.00900990.313535220.20980.5498−0.0003440.211989−0.0002540.2119900.01984580.207747230.10020.55990.0064480.1016000.0079910.1015410.05751270.09673724−0.00770.56970.070414−0.0091130.074630−0.008163−0.316807−0.01350725−0.11080.5800−0.011057−0.123246−0.011042−0.123268−0.055495−0.13010726−0.20770.58770.002748−0.2080680.0039901−0.208016−0.024613−0.214299 Table 10The relative errors and predicted current.DataIoutVoutSDMDDMTDMRelerrorIestRelerrorIestRelerrorIest10.7518−0.2024−0.0000550.7639950.0000040.7638540.0000080.76284420.7598−0.1268−0.0007170.761515−0.0006610.761593−0.0006540.75040030.7593−0.0555−0.0000170.760212−0.0010770.760205−0.0010660.76020440.75930.00240.0003300.7600960.0003990.7600400.00039990.76004450.74780.06130.0001140.7589920.0010430.7589940.00104490.75909960.74680.11520.0011170.7579820.0010380.7579070.00102690.75809970.74480.1645−0.0000140.756994−0.0001110.756923−0.0001230.75706180.74480.21080.0022010.7550280.0009990.7551180.00087190.75611490.74230.25020.0004010.7549950.0003070.7540420.00030590.754044100.74180.28990.0003050.7525440.0002470.7526980.00026590.752690110.74930.3236−0.0011660.750289−0.0011690.750263−0.0009950.750237120.73330.3552−0.0010280.746238−0.0010340.746165−0.0008550.746006130.72530.3840−0.0010490.740095−0.0019060.738862−0.0017230.738741140.71580.41040.0007080.7263010.0009010.7261370.00130650.726991150.70320.4340−0.0005100.705822−0.0003570.7057170.00010990.705391160.66230.45670.000180.6741680.0002630.6741000.00145990.673495170.62980.47510.0016500.6307420.0016300.6307540.00355960.628628180.56080.49370.0005010.5719980.0009930.5710060.00440990.570344190.48580.5099−0.0008300.498258−0.0009040.4984360.00341010.496292200.40080.5232−0.0010410.412353−0.0012170.4124220.00573960.410505210.30330.5365−0.0011290.316192−0.0022030.3161190.00900990.313535220.20980.5498−0.0003440.211989−0.0002540.2119900.01984580.207747230.10020.55990.0064480.1016000.0079910.1015410.05751270.09673724−0.00770.56970.070414−0.0091130.074630−0.008163−0.316807−0.01350725−0.11080.5800−0.011057−0.123246−0.011042−0.123268−0.055495−0.13010726−0.20770.58770.002748−0.2080680.0039901−0.208016−0.024613−0.214299 The relative errors and predicted current. Another consideration is the rate at which the optimal solution is being discovered. An examination of DLBMPA and other baseline algorithms’ performance on 20,000 assessments and population agents from 10 to 100 is carried out in this section to determine SDM, DDM, and PV module characteristics. Figure 6 presents a visual representation of the convergence paths for the SDM, DDM, and TDM models. Take note that the results of SDM, DDM, and TDM are correspondingly shown in the first, second, and third columns of the table. The convergence graph shows that the DLBMPA approach operates best with 60 predators involved. Fig. 6The comparison of the convergence curves for various population sizes in SDM, DDM, and TDM. The comparison of the convergence curves for various population sizes in SDM, DDM, and TDM. Figure 6 compares the convergence of DLBMPA with a variety of other methods. Due to the randomness of the starting location for each method, a reasonable comparison may be made using these charts. Experimental results The I–V characteristics of photovoltaic systems are examined in relation to changes in irradiance and temperature. KC200GT, SW255, and SM55, which are commercially available, are used in our experiment with G = 1000 W/m2 and T = 25 °C; the module specs may be found in Table 11. It should be noted that the I-V data for the KC200GT, SW255, and SM55 PV modules under different temperatures and irradiance levels are presented in Tables A1 to A6. Table 11Standard test data for the module.ParameterCell numberVppmIppm P max VocKIKVIscSM553617.43.155521.71.40− 763.45KC200GT5426.37.6120032.9− 30− 1238.21SW2556031.48.1525537.84− 308.66 Table 11Standard test data for the module.ParameterCell numberVppmIppm P max VocKIKVIscSM553617.43.155521.71.40− 763.45KC200GT5426.37.6120032.9− 30− 1238.21SW2556031.48.1525537.84− 308.66 Standard test data for the module. A photovoltaic current equation depending on irradiance and temperature can be derived from Eq. (1). Table 12 summarizes module estimations. Table 12The results of SDM for RTC France.Parameters \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_s}(\Omega )$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_{shunt}}(\Omega )$$\end{document} Iph (A)Id1(\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document}) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_1}$$\end{document} RMSESM550.234288312.721553.46211.001e − 041.010792150.000321547KC200GT0.2410072760.221708.20210.0013211.056444210.017994510SW2550.1215002811.102108.68440.0154801.218999920.015000010 Table 12The results of SDM for RTC France.Parameters \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_s}(\Omega )$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${R_{shunt}}(\Omega )$$\end{document} Iph (A)Id1(\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mu A$$\end{document}) \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${n_1}$$\end{document} RMSESM550.234288312.721553.46211.001e − 041.010792150.000321547KC200GT0.2410072760.221708.20210.0013211.056444210.017994510SW2550.1215002811.102108.68440.0154801.218999920.015000010 The results of SDM for RTC France. The proposed technique has the potential to achieve fewer RMSEs in both scenarios, as demonstrated. Figures 7 and 8 show the SM55 module’s I–V and P–V graphs for a constant temperature T = 25 °C and irradiance (G = 1000 W/m2) in varied irradiance circumstances. Fig. 7The I–V and P–V curves of the SM55 module with an irradiance level of G = 1000 W/m2. The I–V and P–V curves of the SM55 module with an irradiance level of G = 1000 W/m2. Fig. 8The I–V and P–V plots for the SM55 module at a temperature of 25 °C. The I–V and P–V plots for the SM55 module at a temperature of 25 °C. Figures 9 and 10 exhibit the current–voltage (I–V) and power-voltage (P–V) characteristics of the KC200GT module under varying levels of irradiance while maintaining a constant temperature of 25 °C and a constant irradiance level of 1000 W/m2, respectively. Also, Figs. 11 and 12 show the SW255’s P–V and I–V plots for various situations at a constant temperature, 25 °C, as well as a variety of temperature values at constant 1000 W/m2 irradiances. Fig. 9The P–V (power–voltage) and I–V (current–voltage) graphs of the KC200GT module under an irradiance level of 1000 W/m2. The P–V (power–voltage) and I–V (current–voltage) graphs of the KC200GT module under an irradiance level of 1000 W/m2. Fig. 10The current–voltage (I–V) and power–voltage (P–V) characteristics of the KC200GT module at a temperature of 25 °C. The current–voltage (I–V) and power–voltage (P–V) characteristics of the KC200GT module at a temperature of 25 °C. Fig. 11The current–voltage (I–V) and power-voltage (P–V) characteristics of the SW255 module when G = 1000 W/m2. The current–voltage (I–V) and power-voltage (P–V) characteristics of the SW255 module when G = 1000 W/m2. Fig. 12The current–voltage (I–V) and power–voltage (P–V) characteristics of the SW255 module at a temperature of 25 °C. The current–voltage (I–V) and power–voltage (P–V) characteristics of the SW255 module at a temperature of 25 °C. The output power experiences a drop as a result of increasing temperatures and decreased irradiance. It has been discovered that the power output is inversely and directly proportional to the ambient temperature and irradiance. The values of Id1 and Iph in the results tables may shift with changes in irradiance and temperature. The offset values Rs, Rshunt, and n1 in series resistors are constant for most of the operational range. Iph denotes photocurrent, and it increases with the increase of irradiance. An increase in temperature increases the number of photons that are absorbed and thus leads to an increase in Isd1. Statistics similarly confirmed this occurrence. Time complexity CPU time waste is a crucial metric in the practical analysis of PV characteristics. Computational performance was evaluated by estimating the amount of CPU time needed for each operation. These tests record CPU time in milliseconds as the standard measurement for CPU utilization. Table 13 shows the average processing time for the various algorithms. Compared to other investigation techniques, STOA performs poorly while requiring the least CPU time. In the studies, DLBMPA converges more quickly and precisely than IJAYA, PGJAYA, and WSOA, even though they all need comparable CPU time. The well-known WHHOA approach comes in second in terms of accuracy; however, it is a time-consuming method. Table 13The average computing time for the different algorithms (ms).MethodModulesSDMDDMTDMDLBMPA12.742213.111213.4432PGJAYA1412.788113.279913.5211WHHO1328.814029.110329.4166EHHO1219.743020.076720.4421NLBBO1533.104234.210334.3005AOAHOA1516.143017.120317.5430IJAYA1611.680812.000112.6903STOA1810.011010.812011.0001QRGNMPO1731.201131.700631.7711WSOA1911.243211.301011.7971GTLBOA2112.107012.397512.7922GWO2012.982214.121214.5632 Table 13The average computing time for the different algorithms (ms).MethodModulesSDMDDMTDMDLBMPA12.742213.111213.4432PGJAYA1412.788113.279913.5211WHHO1328.814029.110329.4166EHHO1219.743020.076720.4421NLBBO1533.104234.210334.3005AOAHOA1516.143017.120317.5430IJAYA1611.680812.000112.6903STOA1810.011010.812011.0001QRGNMPO1731.201131.700631.7711WSOA1911.243211.301011.7971GTLBOA2112.107012.397512.7922GWO2012.982214.121214.5632 The average computing time for the different algorithms (ms). Examination of the capability in exploration and exploitation phases Optimization algorithms frequently undergo a dynamic process that consists of two essential phases: exploration and exploitation. Exploration involves the quest for uncharted and undiscovered areas inside the solution space, with the objective of uncovering possibly optimal solutions. Exploitation, however, is about increasing the effort on already mature areas to improve further and perfect already available developments. The degree of success of this optimization process is influenced by how well a balance between these two phases is found. An ideal algorithm, it turns out, has to be able to perform all its actions effectively, including making an easy transition from obsessive exploration to ruthless exploitation. This section presents a thorough investigation of exploration-exploitation behavior in twelve optimization algorithms. Figure 13 depicts the dynamic transition between the exploration and exploitation stages for each method across numerous iterations. The DLBMPA exhibits outstanding performance by outperforming all other algorithms in reaching a well-balanced exploratory-exploitative ratio. Fig. 13The dynamic transition between the exploration and exploitation phases for each method across iterations (a) individual illustration, (b) all algorithms in one figure. The dynamic transition between the exploration and exploitation phases for each method across iterations (a) individual illustration, (b) all algorithms in one figure. The figures illustrate DLBMPA’s gradual and smooth decrease in relation to a ratio of 1, indicating an optimal equilibrium between exploration and exploitation. The performance noted highlighted the ability of the algorithm to navigate the solution space intentionally and to explore new areas that had not been investigated while making use of previously found good areas. The findings highlight the self-adjusting approximation and efficiency of DLBMPA, which enables the so-called exploration-exploitation tradeoff to be obtained. The DLBMPA’s specific features are what sets it apart, and it has been shown to be an effective optimization technique for complex optimization problems in many applications. However, further studies and application of DLBMPA for realistic problems could show its real power and efficiency in solving such optimization tasks. Results discussion The findings indicate that the DLBMPA is more efficient than other available techniques. Specifically, studies have demonstrated its superior performance over techniques like IJAYA, WHHOA, and PGJAYA when it comes to parameter estimation for photovoltaic (PV) models in solar energy. The differences in performance between the proposed algorithm and others are due to several new features incorporated into the design of DLBMPA, which together enhance convergence speed, minimize the complexity of computation, and optimize the exploration-exploitation balance. DLBMPA’s effectiveness is at least partially attributable to its dynamic shift function. Whereas conventional methods employ fixed patterns of exploration and exploitation, DLBMPA transitions internally between Lévy flight (LF) and Brownian walk (BW) patterns of behavior. The design of this change in the optimization method ensures a greater emphasis on LF at the onset of the optimization process, the very initial stage of operation, which encourages a more thorough search across the provided solution space to identify suitable regions. In later iterations, when the approximations of solutions are near optimality and convergence appears plausible, DLBMPA seamlessly merges into BW, tracking the local search. The global LF and local BW transitions ensure that we consistently reach near-optimal solutions faster, avoiding the risk of premature convergence to non-optimal solutions. What this means is that because it is able to adapt, DLBMPA demonstrates both good accuracy and convergence, where, as it is noted, its average RMSE of 9.7 × 10− 4 is better than other techniques. Moreover, DLBMPA has integrated the Fish Aggregating Device (FAD) Mechanism, enhancing the effectiveness of this method. The FAD mechanism injects controlled disturbances into the agents so that they do not settle at local optima. This feature is useful in the particularly challenging cases of high-dimensional, multimodal PV parameter estimation, where many classical methods would become relatively ineffective because they tend to converge to suboptimal points. Periodic FAD interventions enable DLBMPA to achieve great flexibility across different search phases, thereby expanding its bandwidth to search in new areas of the search space when necessary. As demonstrated in the experiments, this mechanism enhances the strength and stability of DLBMPA, enabling it to maintain good performance despite variations in irradiance and temperature, a feat that other methods struggled to achieve under the same conditions. One more critical strength of DLBMPA that is worth mentioning is its lower computational complexity. DLBMPA boasts a radial structure with fewer hyperparameters, making it easier to use and less burdensome than many complex optimization algorithms. In our results, DLBMPA needed about 13 ms on average to complete a single iteration, which is much smaller than the execution time of many relative techniques. This allows DLBMPA to be effective in PV applications where real-time or near-real-time services are needed. Also, the reduced requirements for the parameters allow DLBMPA to be relatively easily applied to various types of PV models, as confirmed by the small RMSE values obtained for the tests of SDM, DDM, and TDM. In a nutshell, the experiments authenticate DLBMPA as the most accurate algorithm for estimating PV parameters. The outcome was the same when the performance of DLBMPA was compared with that of ten algorithms; its accuracy and speed of convergence were superior with a great margin, thus making it ideal for effective use in practical PV systems. Pose the equation with DLBMPA applied for adaptive, accurate, faster, and less computation requirements, and it is proven to be effective and efficient in place of present optimization techniques. Future work goes with the flow as the study could be extended, for instance, by additional real-time applications of PV modeling, by testing DLBMPA in multiobjective or large-scale PV optimization problems, or even by improving the initial parameter settings through the use of experiment methods like the Taguchi approach. To summarize, DLBMPA’s restatement includes overcoming the need for shift-function design and the need for an FAD system. One of the advantages of this design is that it paves the way for better performance in PV parameter estimation without some of the drawbacks experienced by older estimates. This document helps the discipline by demonstrating new domains on DLBMPA as an effective method with a high level of practicality in the modeling of PV system expansion and the use of renewable energy. Conclusion This research proposed the DLBMPA as a novel technique for parameter estimation in PV systems. To overcome the challenges of PV modeling with supposed complexities, DLBMPA proved to be efficient in terms of global search capability, convergence, and robustness. As a result of different tests performed on SDM, DDM, and TDM, DLBMPA was accomplished, and the target accuracy was met with an acceptable average root mean squared error of 9.7 × 10−4, nearly 1000 times lower than the error measure thresholds set for more than ten already recognized optimization algorithms, among them IJAYA, WHHOA, and PGJAYA. Further statistical results approved the existence of the enhancements mentioned above, lending support to DLBMPA in estimating PV parameters. On the other hand, with respect to time, DLBMPA spent on average 13 ms in each iteration, making it have relatively better and faster convergence than the other methods and fit for real-time PV applications. For this algorithm’s evaluation in practice, three commercial PV modules (KC200GT, SW255, and SM55) were used, and the algorithms produced stable photocurrent and saturation current estimates for the parameters investigated despite changes in irradiance and temperature. This proves DLBMPA’s unaffected nature and reliability even in variable conditions, confirming its use in practical modeling of the PV system. The results show that DLBMPA outperforms other benchmark algorithms tested due to the enhanced accuracy and efficiency levels. One limitation of this research is that the effectiveness of DLBMPA can be tested within a confined environmental setting. Future works can concentrate on the progress of DLBMPA to be effective on variable environmental parameters, a larger number of objectives, and organizational complexities. Furthermore, other studies could employ more efficient techniques within the methodological approach, such as the Taguchi method, to enhance the robustness of the design parameters further. Consequently, compared with the existing approaches, DLBMPA seems to possess great promise in view of both accuracy and computational burden, making it practical for PV parameter estimation. Given the evidence of reliability and scope for extensions to DLBMPA, it is expected that DLBMPA will provide a practical strategy for the optimization of renewables while making a significant contribution to the efficiency of solar PV models. Future research directions show promising prospects regarding the expansion of DLBMPA’s applicability range and functionality. For example, the DLBMPA can be enhanced to deal with multicriteria optimization problems in order to achieve several goals, such as maximizing energy efficiency and minimizing costs or effects on the environment. In addition, investigating parallel or distributed variants may increase its potential in terms of scalability and applicability to large PV arrays and intricate configurations of the systems. Further modifications designed and adapted for specific types of PV systems, such as grid-connected or hybrid systems, would make DLBMPA more versatile by adding aspects such as grid stability and integrated storage. To sum up, DLBMPA can be regarded as a molecular and efficient tool for PV parameter estimation as it shows a good tradeoff between accuracy, speed, and robustness. Because of its strong potential for further modifications under many criteria and numerous cases, it stresses its importance for many PV applications, which makes it an essential asset for optimization methods in the modeling and management of renewable energy resources.