Hybrid INC–Fuzzy–LSTM MPPT for Photovoltaic Arrays Under Dynamic Cloud Conditions

Abstract

Photovoltaic (PV) systems operating under rapidly changing cloud conditions experience substantial irradiance fluctuations, partial shading, and multiple local power maxima, reducing the effectiveness of conventional maximum power point tracking (MPPT) techniques. This study proposes a hybrid Incremental Conductance–Fuzzy–Long Short-Term Memory (INC–Fuzzy–LSTM) MPPT controller that improves tracking efficiency, convergence speed, stability, and global maximum power point (GMPP) detection under dynamic irradiance. The controller integrates adaptive Incremental Conductance (INC) for local tracking, a Particle Swarm Optimization (PSO)-tuned fuzzy logic controller for duty-cycle supervision and oscillation reduction, and an LSTM prediction layer for short-term irradiance forecasting. MATLAB/Simulink simulations under four dynamic irradiance scenarios (fast cloud transient, successive cloud events, stochastic fluctuations, and uniform changes) achieved tracking efficiencies of 96.1 %, 95.3 %, 94.8 %, and 97.6 % (mean 96.0 %), mean tracking time of 1.45 s, mean steady-state oscillation of 4.8 W, and GMPP detection rates of 95.8–99.2 %. The proposed controller harvested the highest energy in every scenario. These results indicate that the coordinated integration of adaptive tracking, fuzzy supervisory control, and short-term prediction can improve simulated MPPT performance under rapidly varying irradiance. Because the study is simulation-based, experimental and hardware-in-the-loop validation remain necessary before practical deployment can be claimed.