利益均衡导向的城市综合能源系统多主体分层协同调度控制研究

Cities consume nearly 75% of the world ’s natural resources and 80% of the energy supply, which are the main battlefield of the energy revolution. The contemporary urban energy system is deeply coupled with multiple energy forms such as electricity, cooling and heating within the energy supply-transmission-distribution sides. The city-scale integrated energy system a complex system with heterogeneous energy flows, multi-agent interest correlation, and multi-level interaction effects. The project intends to carry out a multi-subject hierarchical coordinated control study of city-scale urban integrated energy system, with three research topics 1） using a combination of mechanism analysis and data-driven methods to establish multi-energy unified model that includes equipment/network/loads; 2) building up a multi-agent "system-region-user" three-tier scheduling architecture and builds on cooperative/non-cooperative games Interest-balanced multi-agent collaborative management approach; 3) establishing a multi-agent online collaborative optimization control strategy in the energy market environment based on graph neural network and multi-agent deep reinforcement learning technology. This project aims to solve the optimization scheduling problem of city-scale integrated energy system that has the characteristics of high-dimension, none-linear, multi-stage and multi-agent. The research output will bring solid foundation for the promotion of multi-energy city-scale integrated energy system.

城市消费了全球近75%的自然资源和80%的能源供应，是能源革命的主战场。现代城市能源系统在能源“供应-传输-配用”侧深度耦合电气冷热等多种能量形式，是一个异质能流并存、多主体利益关联、多层级交互影响的复杂系统。项目拟开展城市综合能源系统多主体分层协同调控研究，主要包括1）采用机理分析与数据驱动相结合的方法建立包含设备/网络/负荷的多能统一模型；2）构建多主体“系统-区域-用户”三层调度架构，并基于合作/非合作博弈建立利益均衡导向的多主体协同管理方法；3）基于图神经网络和多代理深度强化学习技术，构建能源市场环境下多代理在线协同优化控制策略。项目着力于解决城市综合能源系统这一呈现高维、非线性、多阶段、多主体的复杂系统优化调度难题，研究成果将为多能流城市能源网络的广泛应用提供坚实的理论基础支撑。