Transmission Line Fault Diagnosis Based on Machine Learning

Abstract

In this paper, a faulted phase selection scheme for extracting MPE values of fault transient voltage signals and combining them with CS-SVM for high-voltage transmission lines is proposed. The results show that using MPE to quantify the transient voltage signal in high-voltage transmission lines can fully reflect the fault transient signal characteristics, can be accurately judged within a small time window, identifies faults with high accuracy, is not affected by the fault occurrence location, transition resistance size, and the initial angle state of fault occurrence, and can overcome the defect that the voltage signal has weak sensitivity at the strong power side of the system.