SUMMARY
This paper proposes a multiagent approach to decentralized power system restoration for distribution system networks and presents a comparison of centralized and decentralized systems for power system restoration. Numerous studies have been conducted on power system restoration problems. From the viewpoint of system structures, this research can be divided into two categories: centralized systems and decentralized systems. In this study, we solve a power system restoration problem for a power distribution network by two different methods: mathematical programming (MIP) and the proposed multiagent‐based method (MAS). The proposed multiagent system consists of several distributing substation agents (DSAGs) and load agents (LAGs). An LAG corresponds to the customer load, and a DSAG supplies electricity to the LAG. From the simulation results, it can be seen that the proposed system can reach the right solution by making use of only local information and that the solution quality is better than that of the centralized system (MIP). This means that the proposed multiagent restoration system is a promising approach to larger scale distribution networks.