Agent-based Self-discipline Operation of Integrated Energy System

This paper presents a Self-discipline economic operation method of integrated energy system (IES) considering fault state based on agent. Firstly, various energy supply equipment in the IES are modeled, and the objective function of system economic operation is designed. Then, the Markov decision process is introduced, the economic operation problem of IES is transformed into Markov decision problem, and the agent model of optimal operation is established. Then the agent training based on historical running data is completed by using deep deterministic policy gradient (DDPG) algorithm. Finally, taking the IES of a multi-energy park as an example, the rationality of the output change of each equipment after agent decision under different fault conditions is analyzed.