The line loss management work is closely related to the operation efficiency of a line, the economic benefits of the electric power enterprise, and the safety of power consumption. However, an abnormal relationship between the customer and the transformer will lead to the inaccurate calculation of the line loss in the station area, thus hindering the line loss management work. Therefore, in view of the problems of large workload, high cost, and lack of timeliness of identification results in traditional manual inspection, we first screen abnormal transformers by analyzing the customer–transformer relationship using line loss data collected through power metering sensors. Then, we use the method of trend distance to measure time series similarity, applying the density-based spatial clustering of applications with noise (DBSCAN) clustering algorithm to update and identify any abnormal customer–transformer relationship, and finally verify the design method through experimental simulation analysis.