Permanent and Transient Fault Identification for 10 kV Distribution Network Using Wavelet Analysis and Support Vector Machine

Abstract

Conventional automatic closing may reclose on a permanent fault and cause severe consequences without judging the fault type after a delay. It is proposed to use the fault recording data after the circuit breaker trips to identify the types of faults in real-time. If it is a transient fault, the switch will be closed; then the power supply will be restored. While it is identified to be a permanent fault, the switch will not be closed and wait for maintenance. In this paper, wavelet analysis is adopted to extract real-time features from the fault recording data, and then the support vector machine (SVM) model is used to identify permanent and transient faults, which can avoid blind automatic reclosing.