Wavelet Transform and Stacked Sparse Autoencoder Network Based Location Method of Cable Incipient Fault

Incipient fault of power cables can easily evolve into permanent fault, which endanger the safety and stability of power systems seriously. In order to detect and solve the incipient fault of cables in time, in this paper, a fault location method based on combination of wavelet transform and stacked sparse autoencoder is proposed. Wavelet transform is used for feature extraction of fault signal, and the original signal is decomposed into time domain features of different frequency bands. Stacked Sparse Autoencoder network takes these features as input, training and forecasting, and finally determines the location of fault signal. Experiments show that the method we proposed can locate the incipient fault of power cables quickly and accurately.