Abstract: With the rapid expansion of low-voltage distribution network and the widespread access of distributed energy resources and new power loads, the problems of private connections and unclear station accounts in distribution network in old urban areas have made the operation and management of low-voltage distribution networks and digital transformation face unprecedented challenges. In recent years, with the popularization and enhancement of electricity consumption information acquisition system, it has become possible to realize the digital solution to the physical problems of station topology through the use of collection data. However, limited to the actual collection capacity and on-line rate of electricity consumption information acquisition system, most of the existing results focus on the analysis of the daily load data as a whole, and fail to extract the characteristics of the power consumption behavior contained in the curve data and utilize them in the research. In this paper, we propose a user-transformer relationship identification algorithm based on power step feature sequence matching. First, tree-shaped topology model of low-voltage stations is established. Then, power step features from power curves and composes sequences are extracted, similar peaks and valleys and noise from power curves are subtracted. Then, user-transformer relationship identification of low-voltage stations after sequence matching is carried, and a variety of means are utilized to calibrate and disambiguate the results of the user-transformer relationship identification. Finally, the correctness and efficiency of the proposed method is verified by analyzing an example with 62 meter boxes and 4812 sets of collected data for user-transformer relationship identification.