A Earth Fault Section Identification Method Based on ANN

Abstract

The method to discriminate the fault within or beyond the protective zone according to the signals with high frequency detected in protection has problem with low reliability when the resistance of earth fault is high. The transient voltage with low frequency detected in protection can reflect the size of fault intensity approximately, and the transient voltage with high frequency detected in protection in internal fault is bigger than that in external fault when the size of fault intensity is same. The high frequency transient voltage and low frequency transient voltage detected in protection are chosen as inputs and fault section as outputs for training and testing the neural network for fault identification in the UHVDC transmission line, fault section can be identified when input fault characteristic data to the trained model. A variety of transmission line earth fault situation are simulated by PSCAD, and this earth fault section identification method based ANN shows satisfactory performance, even the fault resistance is high.