Implementation on FPGA of an algorithm for the characterization of solar modules

Abstract

The photovoltaic generator, considered to be the heart of any photovoltaic installation, exhibits sometimes malfunctions during their life cycle, which lead to degradation of the entire photovoltaic installation. Therefore, diagnostic techniques are needed to ensure fault detection, prevent dangerous risks, provide protection and prolong their healthy life. To these ends, this work contributes to the study of the types of degradation of photovoltaic generators, their types of faults and these main diagnostic techniques. This thesis work has been developed into two main parts, the first part "Photovoltaic Generators: Generality, Performances, Productivity, Faults, Diagnosis, Modelling, Characterization, & Identification". This first part is organized in three chapters. While the second part "Artificial Intelligence & Implementation" includes two chapters. Chapter 4: Application of neural networks to the diagnosis of PVG defects. In this chapter, a general description of neural networks and their application to the diagnosis of faults occurring in PVG is presented. It is a network of artificial neural networks, developed to model different types of faults that can appear when operating a PV system in real time. Simulations and experimental tests have been presented, this method shows good results for the modelling and the diagnosis of the healthy and defective photovoltaic system. As a final phase of this thesis, chapter 5 contains a synthesis methodology for the implementation on an FPGA board of one of the neural networks developed. The proposed VHDL description is based on a simple, regular and parallel architecture.