Electric Transmission System Fault Identification Using Modular Artificial Neural Networks for Single Transmission Lines

Abstract

Identifying the fault location is usually the first step in system restoration. Accurate fault location is therefore essential in speeding up repair of faulted components and enhancing system reliability. Diverse fault location methods exist that have different strengths and weaknesses. This paper focuses on using artificial neural networks to classify and locate faults on single two-terminal transmission lines. There has not been sufficient existing work that elaborates how to choose the best neural network structures such as the number of hidden layers and the number of neurons in each layer. This paper aims to study the effects of various ANN structures (number of neurons in a single layer) and identify the most effective neural network structure. Study results based on simulated data are reported, which may provide guidance on designing efficient neural networks for classifying fault types and locating faults on single power transmission lines.