Rastreamento de máxima potência através de redes neurais artificiais aplicado a um sistema fotovoltaico conectado à rede

Abstract

This work presents a study and development of a maximum power point tracking (MPPT) method for a single-phase photovoltaic system that includes an elevator converter and an grid-connected inverter. This tracking strategy is based on the use of multilayer Perceptron artificial neural networks (MLP), in which the information sought at the RNA output is the estimated voltage value to operate at maximum power. To this end, a modelling algorithm for the Kyocera KC200GT panel was developed in order to estimate parameters not informed by the manufacturer. The simulation result returned in a relatively low mean square error, thus showing that the RNA with the proposed approach can be an efficient solution for detecting the maximum power point in several solar irradiation and temperature climatic variations, which showed as an advantageous alternative to the conventional Perturb and Observe method for performing faster tracking.