Performance evaluation of recent metaheuristic optimization algorithms for photovoltaic system parameter extraction

Photovoltaic (PV) systems are getting great deal of attention in renewable energy research. However, the PV system and its power delivery are subject to limiting factors that include temperature, irradiance and wind speed and direction among other factors. Thus, in order to optimize the power delivered by the PV system via current and voltage characteristic curves, a proper modelling of the system is essential. The most common PV system modelling is the electrical circuit based single and double diode solar cell models. The model’s parameters of single and double diode models consist of nodes currents, parasitic resistances, and the diode ideality factor, in which they are subjected to change according to the operation conditions. In addition to the nonlinear nature of these models, the parameter extraction of these models has become a challenging task. This paper presents the application of recent metaheuristic optimization methods namely, Lightning Search Algorithms (LSA), Atom Search Optimization (ASO) and Particle Swarm Optimization (PSO) in order to extract the five parameters of the single diode with a comparison against an experimental setup utilized in this study. In addition, for a fair comparison, algorithm dependent parameters such as number of iteration and population size are fixed for performance assessment. The results show that all the three algorithms provide acceptable results. However, LSA outperform the other algorithms based on standard deviation, consistency and obtained models parameters with an error of 0.0847% compared to the experimental data whereas ASO and PSO have generated a fitness error of 0.157 and 0.146 % respectively. Thus, LSA based model could be used for PV system modelling and maximum power point tracking, after optimizing the execution time.