Reinforcement-Learning-Based Intelligent Maximum Power Point Tracking Control for Wind Energy Conversion Systems

This paper proposes an intelligent maximum power point tracking (MPPT) algorithm for variable-speed wind energy conversion systems (WECSs) based on the reinforcement learning (RL) method. The model-free Q-learning algorithm is used by the controller of the WECS to learn a map from states to optimal control actions online by updating the action values according to the received rewards. The experienced action values are stored in a Q-table, based on which the maximum power points (MPPs) are obtained after a certain period of online learning. The learned MPPs are then used to generate an optimum speed-power curve for fast MPPT control of the WECS. Since RL enables the WECS to learn by directly interacting with the environment, knowledge of wind turbine parameters or wind speed information is not required. The proposed MPPT control algorithm is validated by simulation studies for a 1.5-MW doubly-fed induction generator-based WECS and experimental results for a 200-W permanent-magnet synchronous generator-based WECS emulator.