ERP and DTW-based Transformer-customer Identification

Abstract

The loss management work is closely related to the line’s operation efficiency, the power enterprise’s economic benefits, and electricity consumption safety. However, the strange relationship between the household transformer leads to the inaccurate calculation of the line loss in the station area, thus hindering the line loss management work. Therefore, given the problems of large workload, high cost, and short timeliness of identification results in traditional manual inspection, line loss fluctuation data is used to screen abnormal users of household transformer relationships. Accurate compensation editing distance (ERP) is combined with the dynamic time warping algorithm (DTW) to calculate the similarity of the user voltage curve in the abnormal station area. The SOM clustering algorithm is used to update and identify the household transformer relationship in the abnormal station area. Finally, the correlation analysis and convolutional neural network algorithm are combined to analyze and verify the updated household transformer relationship by using the power outage correlation between the station area and users, which has a specific application value.