Fault location in transmission lines using neural network and wavelet transform

This paper presents a new method of fault location in transmission lines. The method is based on analysis of reflected traveling waves from fault point. Regarding weak frequency response of conventional output of capacitive voltage transformers (CVT), the proposed method receives the travelling waves from PLC output of CVTs, which has a good frequency response for high frequency travelling waves. Received signal is processed by wavelet transform, and signal characteristics are used as input for neural network. After training a neural network, the algorithm estimates the location of fault with reasonable accuracy. The algorithm is independent of the network configuration or length of the line, and is trained once for each voltage level. Numerical studies show the efficacy and accuracy of the algorithm for different configurations.