Topology identification is one of the hot topics of distribution area technology. Topology relation is a universal demand of a power grid. Using the same period energy data of the distribution area to do topology identification is another method and is different from the special topology device and has no additional hardware of topology identification. The characteristic conditions and mathematical combination algorithm of the parent-child relationship of intelligent devices based on Kirchhoff's theorem are studied, and a topology identification algorithm based on cluster analysis is studied. Then the topology identification process from the main switch of a distribution area to the user's electricity meter is realized. The primary data structure and topological data form of intelligent device topology relation are presented. The topology identification technology based on machine learning methods such as cluster analysis and combinatorial optimization algorithm has practical value for the operation and maintenance of a distribution station, and can be used as a reference for the twinning application of distribution data. This work is supported by the Science and Technology Project of Jiangxi Province (No. S2020ZPYFB1256)