A neural-network-based approach for fault classification and faulted phase selection

This paper is concerned with a new approach for fault type classification and faulted phase selection based on artificial neural networks (ANN) to be used for power transmission line protection. The proposed approach is based on a 2-level hierarchical neural network structure. Compared to other architectures, this structure would have a high learning ability and accordingly higher recall accuracy. To reach the corresponding decision, the normalized changes from prefault condition in the instantaneous phase voltages and currents at the relaying point are used. This would lead to an inherent adaptive feature of the approach