Contribution to the detection and localization of faults in the photovoltaic systems

Abstract

With growing concerns about global warming, there is an increasing demand for renewable energy. Among the various types of renewable energies, photovoltaic (PV) energy is the most popular for global electricity production. However, PV power facilities often experience faults that can negatively impact the performance of the PV panels. These faults include hot spots, partial shading, cell aging, cell cracks, short circuits, open circuit faults, and others. An automatic fault detection and diagnosis system is crucial to effectively manage these faults. This thesis addresses two main aspects: PV generator modelling and fault detection and diagnosis. It covers basic principles of photovoltaics, various electrical models for solar cells, and methods of identifying unknown parameters. The research proposes and evaluates a metaheuristic technique for accurately identifying unknown parameters of different three PV models. Additionally, the main contribution of the thesis is a new fault detection method for the PV DC side using Ensemble Learning (EL) techniques. The study achieved advancements in accurate PV generator modelling using metaheuristic approaches, and reliable fault detection using EL techniques