An advanced protection technique based on pattern recognition using wavelet transform in AT feeding system

초록

Distance relay and ΔI fault selection relay are used for main and backup protection relay for protection in AT feeding system in Korea. However, because of the train operation increase, it is very difficult to classify fault and normal operation situation. Especially in the case of high impedance fault caused by the touch of tree, working vehicles, etc, the conventional protection relay can’t detect fault because fault current is very small. In this case, continuous arc can damage electric facility at fault location and stop train operation at the worst. So, we present an advanced technique to classify high impedance fault and temporary overload operation using wavelet transform and pattern recognition technique in this paper. For this, we model AT feeding system, electric train and arcing fault model using PSCAD/EMTDC and simulate the worst operation scenario among the various operation and high impedance fault situation. And we extract characteristics of high impedance fault and worst operation conditions using wavelet transform and detect a fault using adaptive network fuzzy inference system based on pattern recognition. Finally we analyze fault detection performance of the proposed technique compared with the conventional distance relay.