Résumé:
This thesis investigates advanced control strategies for photovoltaic (PV) systems to
address challenges arising from climate variations and operational conditions in both
stand-alone and grid-connected configurations. The research is divided into four main
chapters: an overview of PV generators, power electronics design, intelligent control
techniques, and metaheuristic algorithms for maximum power point tracking (MPPT)
under partial shading conditions (PSC). The study begins by analyzing the behavior of
PV generators and their integration with DC/DC converters. Intelligent control methods,
including Perturb and Observe (P&O) and Adaptive Neuro-Fuzzy Inference System
(ANFIS), are employed to enhance MPPT efficiency under varying irradiance levels, with
ANFIS demonstrating superior performance. Hybrid approaches combining Fuzzy Logic
Control (FLC) and Artificial Neural Networks (ANN) are compared against Incremental
Conductance (IC) methods for grid-connected systems, highlighting the hybrid
controller’s ability to track maximum power dynamically without prior PV system
information. For fault diagnosis, a neural network (NN) model achieves high accuracy in
detecting PV faults. Lastly, metaheuristic MPPT controllers—Grey Wolf Optimization
(GWO), PSO, and Adaptive PSO (APSO)—are evaluated under PSC. The APSO
algorithm excels in achieving fast and efficient MPPT. Simulations conducted using
MATLAB/Simulink validate the effectiveness of the proposed methods, showcasing their
potential for real-world applications in renewable energy systems.
