Détection et localisation des défauts d’un système PV par les techniques soft computing

Marah, BACHA (2025) Détection et localisation des défauts d’un système PV par les techniques soft computing. Doctoral thesis, Faculté des Sciences et de la technologie.

Text Marah__BACHA Thesis.pdf Download (9MB)

Abstract

Ensuring the reliability and efficiency of photovoltaic (PV) systems requires robust fault detection and diagnosis methods. This thesis presents a comprehensive study on the detection, classification, and localization of PV faults using Artificial Neural Networks (ANN) and Fuzzy Logic (FL) techniques, validated through both simulation and experimental approaches. In the first phase, a simulation model was developed in Matlab/Simulink® to describe the system behavior for both healthy and faulty operations. To deal with this concern, a Matlab/Simulink® co-simulation strategy is developed to elaborate a trusted simulation model. This model requires the use of the One Diode Model (ODM) electrical parameters. For this, an efficient strategy, based on the War Strategy Optimization (WSO) algorithm, is applied to identify the ODM parameters. Finally, the ODM identified parameters are used to elaborate an efficient strategy of maximum power point (MPP) estimation. The efficiency of the developed strategies is experimentally evaluated by using real measured data. In the first phase, the thresholding method and FL classifier demonstrated high fault detection capabilities to diagnose eight types of faults occurring in PV cells, achieving approximately 100% accuracy in the simulation and experimental tests. In the second phase, an Artificial Neural Network (ANN)-based fault detection and classification approach was successfully implemented and validated through both simulation and experimental analysis. Five distinct single- and multi-fault types, including partial shading, open circuit and bypass diode failures, were applied to a PV module through the dSPACE DS1104 controller, confirmed the model’s robustness and reliability. The ANN-based method achieved an impressive classification accuracy of 99.7%, proving its efficiency in detecting PV faults under varying conditions. In the third phase, a comprehensive study has substantiated that both Artificial Neural Networks (ANN) and Fuzzy Logic (FL) are capable of detecting and classifying all single- and multi-fault types effectively. However, when moving from simulation to experimental tests using the dSPACE DS1104 platform, the results unequivocally showcased the superiority of the ANN classifier over the FL classifier. The ANN classifier exhibited superior accuracy (99.6%) and faster fault classification compared to the FL classifier (99.2%) in real-time conditions. The findings of this research highlight the ANN-based approach as an efficient solution for PV fault diagnosis, offering enhanced accuracy and faster processing. These results underscore the potential of integrating ANN techniques into real-time monitoring systems to improve the performance, reliability, and safety of photovoltaic installations.

Actions (login required)

View Item