A novel wavelet-neural network method for fault location analysis on transmission lines

This paper presents a technique, based on discrete wavelet transform (DWT) and back-propagation neural network (BPNN), to find the fault location on single circuit transmission lines. The proposed method has been applied to IEEE 9-bus test system. In order to go through this, MATLAB was used to apply DWT on the signal of fault currents of all the existed generators. The Daubechies Four (db4) mother wavelet is employed to decompose the high-frequency component of fault signals. The norm of detail coefficients of five decomposition levels for all fault current signals was selected as input pattern for the training process of a BPNN. The obtained results show that trained BPNN can be used as a proper tool to detect the location as well as the type of the occurred faults on the system, with a reasonable accuracy.