浙江电力2025，Vol.44Issue(9)：46-57,12.DOI:10.19585/j.zjdl.202509005
基于多智能体分层强化学习的多园区综合能源系统优化运行
Optimal operation of MPIESs based on hierarchical multi-agent reinforcement learning
摘要
Abstract
With the rapid development of multi-park integrated energy systems(MPIESs),individual parks,as in-dependent stakeholders,exhibit differing interests and face challenges such as limited access to global information,a strong emphasis on privacy protection,and the need for autonomy in collaborative processes.To address these is-sues,this paper proposes an optimal operation model for MPIESs based on an energy-sharing business model and hi-erarchical multi-agent reinforcement learning.First,an energy-sharing business model is designed using a business model canvas,and a collaboration framework based on hierarchical multi-agent reinforcement learning is developed to accommodate the MPIES featuring multiple agents.In this framework,the upper-layer coordinating agents opti-mizes global resources and coordinates energy sharing,while the lower-layer executing agents manages energy op-erations and makes energy-sharing decisions.Second,a multi-agent collaborative optimization model for MPIESs is developed and solved using the alternating direction method of multipliers(ADMM)to achieve decentralized global optimization.Simulation results demonstrate that the proposed framework significantly enhances the operational eco-nomic benefits and resource utilization efficiency of MPIESs,reduces dependence on external energy markets,and enhances renewable energy integration.关键词
多园区综合能源系统/能源共享/协作优化/多智能体/分布式优化Key words
MPIES/energy sharing/collaboration optimization/multi-agent/distributed optimization引用本文复制引用
李忠凡,陈曦,黄海涛..基于多智能体分层强化学习的多园区综合能源系统优化运行[J].浙江电力,2025,44(9):46-57,12.基金项目
国家自然科学基金(U2066214) （U2066214）
