The continuous development of the economy has made the scale of the lowvoltage distribution network increasingly large, but at the same time exposes some problems in the management of the line loss in the transformer areas. The fault of transformermeter relationship in transformer areas is one of the main causes of abnormal line loss. In order to make up for the shortcomings of maintenance personnel who need to check the methods of all users on a casebycase basis, a datadriven method for verifying the transformermeter relationship of the district is proposed. The model uses the historical data of power supply and power consumption in the transformer area, and optimizes the daily line loss rate of the transformer area as the optimization target, and adopts the method of rolling optimization under the longterm scale, and finally gives the user who is suspected to have a transformer meter relationship error.Finally, by using two actual case studies, the result shows that the model can give a set of users with suspected file errors, and realizes the efficiency of the transformermeter relationship verification.