ANN Fault Location Algorithms with High Speed Training

This paper proposes Artificial Neural Networks (ANN) based fault location algorithms with high speed training for transmission lines. Inputs of the algorithms are one cycle post fault voltage/current measurements taken from either one or two terminal of the transmission line. The impact of the input type is tested with the phase and the symmetrical component values of the measurements. Evaluation of algorithms are achieved on a 400 kV transmission line for various fault types, fault locations and fault resistances. Training set is created with various cases for the proposed test scenario. The algorithms have been tested at extreme conditions; close to reference point, close to the end of the line, low fault resistance and high fault resistance. The proposed algorithms have resulted in a fairly good fault location estimation even with a limited number of training data for all cases.