Fault Location Method of Underground Cable Based on Wavelet Analysis and Neural Network

In order to solve problems of poor reliability and accuracy of existing fault location methods of underground cable, the paper introduced a fault location method of underground cable based on wavelet analysis and neural network, and compared performance of BP neural network and RBF neural network used in the method. The method uses 3B-spline semi-orthogonal wavelet to do wavelet transformation for transient-state zero-sequence current so as to get modulus maxima of transient-state zero-sequence current in specific frequency bands. The modulus maxima is taken as inputting signals of neural network, and realizes fault location according to mapping relationship between the modulus maxima and position of fault point. The simulation results showed that the method can realize fault location of underground cable, and the method with RBP neural network is better than BF neural network in location error and network training.