Privacy-Preserving Distributed Dispatch for Integrated Electricity and Heat Systems: A Multi-agent PPO Method
G. F. Xu, Qiuwei Wu, Zhenjia Lin, Wai Kin Chan, Ye Guo, Zepeng Li · 2024
Considering that energy suppliers within the Integrated Electricity and Heat System (IEHS) typically provide only partial information to other suppliers for privacy protection, traditional centralized scheduling methods are no longer applicable. This paper proposes a method based on multi-agent deep reinforcement learning to address the distributed dispatch problem within the IEHS. To transform the optimization problem of the IEHS into a reinforcement learning model in a multi-agent environment, the IEHS is divided according to different stakeholders into electrical system agents and heat system agents, thereby establishing a partially observable Markov decision process. Subsequently, the multi-agent proximal policy optimization algorithm is utilized to solve this problem, creating a training and solution framework suitable for the IEHS that supports centralized training and decentralized execution.