Deep reinforcement learning for MPPT control under uniform and partial shading conditions

Maximum Power Point Tracking, MPPT ensures that PV systems operate at maximum output power that is particularly effective under uniform solar irradiation. However, conventional MPPT techniques become less effective under partial shading conditions . This is due to the presence of multiple local maxima on the power–voltage curve. In this study, MPPT is performed using a Deep Deterministic Policy Gradient, DDPG, deep reinforcement learning, DRL controller. Its performance is evaluated under varying operating conditions. The proposed DDPG – DRL agent learns to optimize the PV output under dynamically changing irradiance and load conditions. Simulation results show that the proposed DDPG – DRL controller effectively handles the non-linear variations in solar irradiance, even under partial shading conditions.