An analysis on the performance of metaheuristic algorithms for the estimation of parameters in solar cell models

Abstract

Alternative energies will be the engine that moves the world in the future; therefore, researchers focus their attention on the problems related to designing and properly exploiting this type of energy. Artificial intelligence methods are explored as tools to improve different branches of alternative energies. In this context, metaheuristic algorithms (MA) are optimization techniques that have been explored in the field of solar cell design. This article analyses the performance of different MAs for modeling solar cells by estimating the parameters of photovoltaic (PV) diode models. The MA used are part of a recent generation of techniques that include innovative processes that permits a proper selection of the optimization operators in the iterative search; such algorithms are the Artificial Hummingbird Algorithm (AHA), Aquila Optimizer (AO), African Vulture Optimization Algorithm (AVOA), Equilibrium Optimizer (EO), Golden Eagle Optimizer (GEO), Side Blotched Lizard Algorithm (SBLA), Social Network Search (SNS), and Reptile Search Algorithm (RSA). We implemented them to optimize the parameters of different models of solar cells with arrays of one, two, and three diodes. The experimental analysis performed on an R.T.C. France solar cell demonstrates which methods are the most suitable for the identification of unknown parameters in three PV models named single diode solar cell model (SDSCM), double diode solar cell model (DDSCM), and triple diode solar cell model (TDSCM). The results were statistically analyzed with Voltage–Current and Voltage–Power tests, comparisons of the best-optimized parameters, convergence plots, four error types such as Root Mean Square Error (RMSE), Absolute Error (AE), Mean Absolute Error (MAE), and Normalized Root Mean Square Error (NRMSE), a run-time analysis and non-parametric tests. The study converges in comparing and defining which of the implemented MA’s is the best alternative for the design of solar cells, concluding that some tested methods, such as AHA, can improve the efficiency of this class of devices; with RMSE values of 0.000986 for SDSCM, 0.00098353 for DDSCM, and 0.0009855 for TDSCM and the fourth-best average execution time according to the overall ranking.