Distributed Cluster Regulation Strategy of Multipark Integrated Energy System Using Multilayer Deep Q Learning

Abstract

The power system is evolving from a single energy system to an integrated energy system. In order to further improve the power generation and consumption balance capacity of the park integrated energy system (PIES), the park integrated energy system is gradually transitioning from the single park energy system operation mode to the multipark energy system operation mode. The design of multipark integrated energy system (MPIES) collaborative control strategy will become an important part to improve the power generation and consumption balance ability of the integrated energy system. In order to fully tap the regulation capacity of each PIES, we propose a coordinated control strategy for the integrated energy system in multiple parks considering the flexible substitution interval of multiple types of energy. Firstly, we analyze the influence of the types of regulation resources and the regulation incentive mechanism of the PIES on the regulation flexible range of the PIES. Then, based on the Markov decision process, a distributed cluster regulation model of MPIES considering regulation demand and regulation flexible interval is established. Finally, using multilayer deep Q networks (MLDQN), the distributed cluster regulation optimization algorithm of MPIES is given. The simulation results show that the proposed method can coordinate the regulation ability of each park integrated energy system in the MPIES, give full play to the large-scale advantage of the interconnection of the park integrated energy system, and improve the overall stability of the multipark integrated energy system.