The application of artificial neural networks to transmission line fault detection and diagnosis

Abstract

Fault Detection on transmission lines forms an important part of monitoring the health of the power plant and is an indicator of when potential faults can lead to catastrophic failure of equipment. This research analyses the early detection of generator, transmission line faults and also researches methods of fault detection via the application of Artificial Neural Network techniques. The monitoring of the generator voltages and currents, of transmission line performance parameters forms an important monitoring criterion of large power systems. Failures lead to system down time, damage to equipment and it presents a high risk to the integrity of the power system, and affects the operability and reliability of the network. This dissertation therefore deals with fault detection on the Eskom transmission lines from a simulation perspective. Electrical faults have always been a constant source of conflict between transmission lines and power consumers. This dissertation presents the application faults detection on the transmission lines using Artificial Neural Networks. The ANN is used to model and to predict the occurrence of a transmission line fault, and classifies faults according to its transient characteristics. Results show that the ANN can be used to accurately identify and to classify faults, given accurate problem set-up and training. The major contribution of the dissertation is the application of ANNs to predict faults on the transmission lines