Fast and accurate fault location by extreme learning machine in a series compensated transmission line

This paper presents an improved hybrid approach for fault location in a series compensated transmission line. The proposed method uses one cycle post fault current and voltage samples. Thereafter features of faulty signal are extracted by wavelet transform. Best features are then selected by genetic algorithm based feature selection method and are fed as input to the extreme learning machine for fault location. The performance of the proposed method has been evaluated on a 300 km, 400 kV transmission line with thyristor controlled series capacitor placed at the middle. The proposed scheme has been tested for a wide variety of operating conditions like different fault inception angle, fault resistance, fault location and type of fault. Simulation result shows that the proposed method is quite fast and accurate for fault location in a series compensated transmission line.