Agent-Based Optimal Cooperative Operation of Multi-energy System

With the rapid development of distributed renewable energy, the traditional energy node model has been difficult to adapt to the energy information coupling system. Aiming at the problem that traditional energy node model is difficult to combine information flow and energy flow to establish a rapid response mechanism, this paper uses machine learning to build an agent model based on data and model, which can balance disturbance automatically and tend to run well automatically. Firstly, the scheduling objective and energy converter of the multi-energy system (MES) are modeled, then the agent model is introduced, the observation variables and action space of agent are defined, and the reward function is constructed. After that, the solution process of DDPG algorithm is introduced, and the parameter of DDPG algorithm is completed. Finally, an example is given to verify the effectiveness of the proposed method.