Real-Time Scheduling Method Based on Deep Reinforcement Learning for a Hybrid Wind-Solar-Storage Energy Generation System

With the large-scale penetration of renewable energy in recent years, the volatility and randomness have posed huge challenges to the optimal operation of the power systems. Meanwhile, because of the model uncertainty, parameter uncertainty and time coupling, it is very difficult to develop an universal real-time energy scheduling method for various situations. To cope with the challenges, this paper proposes a novel real-time scheduling method for a hybrid wind-solar-storage energy generation system. The real-time scheduling problem is firstly formulated as a Markov decision process (MDP), and then a deep reinforcement learning (DRL) method based on deep deterministic policy gradient (DDPG) algorithm is proposed to solve this problem. Different from the traditional model-based methods, the proposed method is model-free and does not need to know or forecast the uncertain information (e.g., renewable energy generation, electricity demand, etc.). The proposed method can learn from a large amount of past data to make optimal decisions. Finally, simulation results based on various scenarios verify the effectiveness of the proposed method in improving renewable energy consumption, reducing the total generation cost, and smoothing power fluctuation, etc.