ANN applications in fault locators

Recent developments indicate that Artificial Neural Networks (ANNs) may be appropriate for assisting dispatchers in operating electric power systems. The fault location algorithm being a key element in the digital relay for power transmission line protection, this paper discusses the potential applicability of ANN techniques for determination of fault location and fault resistance on EHV transmission lines with remote end in-feed. Most of the applications make use of the conventional Multi Layer Perceptron (MLP) model based on back propagation algorithm. However, this model suffers from the problem of slow learning rate. A modified ANN learning technique for fault location and fault resistance determination is presented in this paper. A reasonably small NN is built automatically without guessing the size, depth, and connectivity pattern of the NN in advance. Results of study on a 400 kV transmission line are presented for illustration purposes. Performance of the modified ANN is compared with the analytical algorithms and conventional MLP algorithm for different combinations of pre-fault loading condition, fault resistance and fault location. The results are found to be encouraging