Abstract: Due to the great advantage in regulating the peak level in wet season and compensating the storage deficit in dry season, the hybrid pumped storage reservoir gets more attention from researchers in the scientific world. In this paper, the mid-long term stochastic optimization model for hybrid pumped storage reservoir is established to obtain the maximum energy generation based on Reinforcement Learning Algorithm. The case study simulates the designed model and decide the operational policies to one hybrid pumped storage reservoir.