Fault Classification in a Transmission Line Using Levenberg–Marquardt Algorithm Based Artificial Neural Network

Abstract

The main objective of the power system is to supply reliable and quality electricity to all consumers. In this paper, the main focus of the author is to classify all types of faults, namely phase to ground, phase to phase, three-phase fault, and double line to ground faults that may occur at different fault locations and involve varying fault impedances in the power system using artificial neural networks (ANNs). Owing to the advantages of an artificial neural network to map nonlinearity in the data, to learn from examples and to generalize the pattern classification, ANN framework under supervised learning is implemented as a fault classifier. The proposed methodology includes extraction of features from phase voltages and currents obtained under normal and faulty conditions for different fault locations and fault impedances. The learning of feed forward ANN-based fault classifier is carried out using Levenberg–Marquardt algorithm for training the data obtained for IEEE 14 bus system.