PERFORMANCE COMPARISON OF MAXIMUM POWER POINT TRACKING (MPPT) ALGORITHMS FOR SOLAR PV SYSTEMS IN VARYING ENVIRONMENTAL CONDITIONS

Abstract

Maximum power point tracking (MPPT) remains a critical determinant of photovoltaic (PV) energy yield under real-world variability in irradiance, temperature, and partial shading. This thesis develops a unified, validated MATLAB/Simulink framework to enable apples-to-apples comparison of representative MPPT families on a common 10.224 kW PV plant (Amerisolar AS-6M24-170, 4S15P) interfaced to a buck-boost converter at 50 kHz with explicit parasitics. The platform standardizes plant and converter models, controller I/O, disturbance profiles (STC, fast irradiance steps, temperature sweeps, multi-peak partial shading, measurement noise), and a post-processing pipeline for tracking efficiency, settling behavior, ripple/overshoot, and energy loss, so that performance differences are attributable to control design rather than modelling artefacts. Controllers assessed include P&O, Incremental Conductance (INC), Fuzzy-Logic Control (FLC), Artificial Neural Networks (ANN with gradient-descent and Levenberg-Marquardt training), and Grey Wolf Optimizer (GWO). Results reveal a consistent speed-regulation trade-off: intelligent methods deliver superior steady-state quality and damping, with FLC achieving fast recovery from transients and ANN-LM attaining the lowest ripple once locked, while GWO reliably escapes local maxima and attains near-GMPP operation under partial shading. Among classical techniques, INC generally exceeds P&O in accuracy while retaining simplicity. Implementation analysis shows ANN-LM with the shortest runtime, FLC with the lowest memory footprint, and GWO with the highest memory demand, informing embedded feasibility. The study’s contributions are (i) a transparent, reusable benchmarking testbed and (ii) evidence-based guidance that maps controllers to operating regimes, motivating an event-driven hybrid in which a global searcher is engaged episodically and hands off to a low-ripple regulator for steady operation. The framework and findings provide a practical basis for controller selection and future hardware-in-the loop validation.