Contribution à la modélisation et à la simulation des dispositifs photovoltaïques par méthodes évolutionnaires

Abstract

The extraction of electrical parameters of photovoltaic devices is a crucial operation for their modeling, simulation and optimization. Several models have been established to predict their behavior and since the 1960s a large number of methods have been proposed to extract their parameters. In the last two decades, metaheuristic methods, and mainly evolutionary methods (EAs), have been intensively used thanks to their unique advantages including the non-requirement of an initial point, the simultaneous estimation of all parameters and the ability to escape local minimums. In this thesis, we first demonstrate the superiority of EAs then we study the effect of the crossover operator on genetic algorithms to estimate the parameters of a solar cell in a noisy environment. The study, carried out according to the size of the population, the number of generations and the noise level, shows that GAs with the blend operator compete with the most efficient EAs. The GAs are then applied to two solar modules to study the effect of temperature and illumination on the estimated parameters. The study shows that the variation as a function of temperature follows the theory for strong illuminations. For variation as a function of illumination, the behavior has not been previously predicted.