An automatic diagnosis method of power consumption anomaly of station users based on the k-medoids clustering algorithm

Abstract

Abstract Obtaining reliable data on electricity consumption can be difficult due to faulty or inaccurate data acquisition equipment. Therefore, a k-medoids clustering algorithm is used to design an automatic diagnosis method of power consumption anomaly. The K-Medoids algorithm was used to cluster the power consumption data of users in the Taiwan area. The data dimensions suitable for automatic diagnosis are screened by the ADF method. Based on this, the power consumption anomaly of the distribution network station area is automatically diagnosed, and the marked power consumption behavior data characteristics are checked step by step to realize the automatic power consumption anomaly diagnosis of station area users. The experimental results show that the K-medoids clustering algorithm can reasonably avoid the influence of transient abnormal data caused by isolated points on the automatic anomaly diagnosis results. For different types of abnormal automatic diagnosis rate of more than 98.6%, can accurately diagnose the abnormal power consumption of users in the station area.