Neural network based multi-parameter feeder protective relay in traction supply system

With the development of electrified railways, higher demands on feeder protective relaying have been put forward. Although conventional feeder protection relays generally work well, they cannot meet these new demands. A neural net-based multi-parameter feeder protective relaying approach is proposed in this paper. Because this novel approach can express the inherent relations between the signal features, it is suitable for the adaptive protective relaying and the detection of complex faults. Computer simulation results show that this novel approach is both effective and satisfactory