Backtracking search algorithm with Levy flight for estimating parameters of photovoltaic models

An accurate mathematical model plays an important role for simulation, evaluation and optimization of photovoltaic (PV) models. The characteristic current equations describing the PV models are implicit, nonlinear and transcendental. Given the features of the characteristic current equations, traditional optimization algorithms are usually easy to converge to local optimal solutions. Thus using metaheuristic methods called modern optimization algorithms to estimate parameters of PV models has been a research hotspot in recent years. Although many metaheuristic methods have been employed to solve this problem, it is still necessary for researchers to propose new optimization algorithms to obtain more accuracy and reliability solutions. This paper presents a new metaheuristic algorithm called backtracking search algorithm with Levy flight (LFBSA) to estimate the parameters of PV models. Compared with the basic backtracking search algorithm (BSA), LFBSA has the following two remarkable features. Firstly, an information sharing mechanism with Levy flight is built to enhance population diversity. Secondly, mutation operator based on the hunting mechanism of grey wolves is introduced to increase the chance of LFBSA to escape from local minima. LFBSA is used to estimate parameters of three different PV models. Experimental results show the proposed LFBSA is superior to BSA and the other compared algorithms in terms of accuracy and reliability.