A multi-agent distributed collaborative optimization method for urban integrated energy systems with enhanced resilience

With the frequent occurrence of extreme climate events, such as typhoons, lightning, extreme cold, and extreme heat, the resilience of urban integrated energy systems (UIES) has become a core issue concerning the safe operation of cities. This paper proposes a multi-agent distributed collaborative optimization method (MADCO-R) aimed at enhancing the resilience of the system in the face of fault disturbances, with the goal of achieving rapid response and efficient recovery. Firstly, a refined model of UIES considering multiple energy flows (electricity, heat, and gas) was constructed, and resilience quantification indicators were introduced. Secondly, a multi-agent architecture based on physical coupling boundary decomposition was designed, dividing the global system into multiple geographical or functionally autonomous regional energy agents (REA) and a system coordination agent (SCA), clarifying the core logic of autonomous decision-making and distributed regulation of the agents. On this basis, a distributed optimization framework integrating an improved alternating direction method of multipliers (ADMM) and a consensus algorithm was proposed, providing a quantitative coordination basis for the distributed regulation of agents, and achieving multi-energy complementarity and global resilience optimization under the premise of privacy protection. Finally, the proposed MADCO-R method was verified through a UIES test case that couples an improved IEEE 33-node distribution system, a 20-node thermal network, and a 14-node natural gas network. The test results show that, compared with centralized optimization and traditional distributed methods, the proposed MADCO-R method can effectively enhance the resilience level of the system by 10% - 25%, reduce the reduction of critical loads, and significantly enhance the robustness and scalability of the optimization process.