Optimal scheduling of virtual power plant based on Soft Actor-Critic algorithm

Aiming at the problem that distributed power sources are difficult to be centrally scheduled and optimized, a virtual power plant (VPP) optimization scheduling method based on deep reinforcement learning is proposed. The method can adaptively adapt to the uncertainty of source load and give real-time optimal scheduling strategies in the presence of source load fluctuations. First, the system framework of VPP is constructed, and the optimal scheduling model of VPP is established with the objective function of minimum daily operating cost, and various operating condition constraints are practically considered. Then, the basic principle and process of deep reinforcement learning are introduced, and the constructed VPP optimal scheduling model is transformed into a deep reinforcement learning model, and the state space, action space and reward function are reasonably designed. Following that, Soft Actor-Critic algorithm (SAC) is used for offline training, and the trained model is utilized for online decision making. Finally, the simulation is carried out in an area as an example, and the results show that the VPP optimal scheduling strategy proposed in this paper can effectively reduce the operating cost of VPP and ensure the safe operation of the grid after grid connection.