Research on Intelligent Protection Technology for Distribution Network with Distributed Generation

Connecting large-scale distributed generation to a distribution network is complex as it is difficult to set the current protection limit within a traditional distribution network. This paper proposes a distributed protection and control scheme based on reinforcement learning for a distribution network connected to a distributed generation system. This research introduces the idea of transforming this distribution network protection problem into a multi-agent reinforcement learning problem. The concrete construction method of the reinforcement learning environment and deep Q-learning network is described. The algorithm training process is presented and the algorithm training and verification environment is built.