Power Cable Fault Diagnosis Based on Wavelet Analysis and Neural Network

Abstract

With the widespread use of cables, the problem of power system fault diagnosis is becoming more and more serious. As we all know, the sudden blackout caused by cable fault will bring serious threat to the life and property safety of users, and even cause adverse social impact. Avoiding losses caused by cable failures is popular. To diagnose the cable faults is vital to guarantee the safe and steady operation of power transmission line. The combination of wavelet analysis and neural network is adopted as the fault diagnosis method to realize the accurate identification. Wavelet packet decomposition is used for feature extraction of cable fault signals which are input vectors after normalization processing. Radial basis function (RBF) network structure is built and relevant practice and test of cable fault diagnosis are conducted select 8 sets of samples for testing. By selecting different failover resistance values, the target output of the first four groups is 0.9, and the actual output is also 0.9; the target output of the last four groups is 0.1, the actual output is also 0.1, and the target output and actual output are The error between the two is basically zero, which also indicates that the RBF network has good fault discrimination. According to the test result, it shows that this method can be effectively achieved in cable faults diagnosis.