Neural network approach to power transmission line fault classification

This paper presents a new solution to fault classification of high voltage transmission lines and shows its effectiveness in digital simulation on a realistic 500 kV power system. The scheme is based on backpropagation and Kohonen neural networks and a comparison between them is made. The Electromagnetic Transients Program (EMTP) is used to obtain fault patterns for the training and testing of neural networks. Feature selection, feature extraction and signal procession are studied and a fast, reliable fault classifier is obtained.