The low-voltage substation area has long faced the problems of lack of topology and unclear household transformer relationship. In recent years, the refined management and control of distribution network require clear household transformer relationship. In this paper, a household transformer relationship identification scheme based on data mining technology is proposed. Firstly, based on the similar characteristics of voltage fluctuation of nodes in the neighbourhood area, the suspected nodes that do not belong to the target station area are clustered by DBSCAN algorithm; Secondly, based on the power similarity of upstream and downstream equipment, confirm whether the suspected node belongs to the target station area. The Apriori algorithm is used to generate the neighbourhood area dependent node set that meets the constraints, and then the cosine similarity discrimination is used to obtain the most possible household transformer dependent result. Finally, the effectiveness of the proposed algorithm is verified by comparing the results of an actual neighbourhood area network.