Research on Capacity Configuration Optimization of Multi-Energy Complementary System Using Deep Reinforce Learning

The output power of wind, solar, and hydro energy in a multi-energy complementary system (MECS) with the heating system exhibits certain fluctuations. Gas power generation and battery can reduce these problems. However, relying solely on the experience of designers to determine the capacity configuration is challenging, as it may compromise the system's safety and result in wasteful investments. To address these issues, the capacity configuration optimization problem of the MECS can be formulated as a multi-objective optimization problem. This paper proposes a MECS capacity optimization method based on deep reinforcement learning (DRL), specifically the deep deterministic strategy gradient (DDPG) algorithm. By transforming the multi-objective optimization problem into a Markov decision process, this approach effectively resolves it. The primary objective of the optimization is to ensure the stability of the system's power supply, while the secondary objective is to minimize the economic cost of the system. To evaluate the proposed method, simulations were conducted based on the characteristic curves of energy generation. The results demonstrate that utilizing the DDPG algorithm enables the rapid determination of the optimal capacity configuration for wind, solar, and battery. This approach improves the stability and economic efficiency of the multi-energy complementary system, while utilizing excess electrical energy for heating through heat pumps, greatly improving energy utilization efficiency.