QGFSMC: a Q-learning-assisted fuzzy integral sliding MPPT controller for PV power boost conversion systems

Abstract

With increasing renewable-energy demand, efficient maximum power point tracking (MPPT) remains essential for photovoltaic (PV) systems operating under nonlinear and varying conditions. This paper proposes a hybrid Q-learning-based fuzzy sliding mode controller (QGFSMC), integrating Q-learning adaptation, fuzzy decision-making, and gain-scheduled sliding mode control to improve dynamic tracking and disturbance rejection. The coordinated framework provides rapid convergence while maintaining robust operation under irradiance variations and load disturbances. Simulation results demonstrate superior performance compared with P&O, incremental conductance (INC), and model predictive control (MPC). QGFSMC achieved an effective startup rise time of approximately 0.05s, startup settling time of 0.06s, and post-disturbance settling time of approximately 0.02s following disturbances at 𝑡 = 1s and 𝑡 = 1.5s. Furthermore, QGFSMC maintained the highest minimum transient PV power, demonstrating improved disturbance rejection, faster recovery, and enhanced dynamic MPPT performance under varying operating conditions.