Hierarchical Collaborative Optimization of Shared Energy Storage With Co-Generation Based on Deep Reinforcement Learning and P2P Network Game Theory

With the large-scale integration of massive, dispersed, and diverse electric heating flexibility resources into communities, traditional physical energy storage devices are difficult to apply on a large scale due to high construction costs. Electricity building suppliers (EBPs) are an effective way to participate in energy management and low-carbon economic operation of demand-side energy systems. Firstly, this article takes a co-generation type shared energy storage system consisting of high-temperature solid heat storage, waste heat boilers, and steam turbines as a typical case. Based on explaining the basic principles of system operation, the pricing mechanism and optimal load distribution mechanism of community-shared energy storage on the distribution side are studied. Secondly, a double-layer network market model consisting of a P2P energy trading network composed of renewable energy stations and user communities containing co-generation shared energy storage is proposed, and a reinforcement learning Nash bargaining combined ESP and multi-community double-layer energy management model is established. The outer layer optimizes the optimal pricing strategy between the ESP and EBPs alliance through reinforcement learning models, and the inner layer optimizes the “electricity heat” P2P trading strategy within the EBPs alliance through Nash bargaining models. Iterative optimization of inner and outer models. Finally, a case study was conducted on typical systems to analyze the optimal energy management strategy of MEBPs in P2P mode. The results showed that the proposed energy management method can effectively schedule electric heating, achieve maximum system revenue, and reduce carbon emissions.