This study has as objective the application of Artificial Neural Networks (ANNs) theory as pattern classifiers. The implemented neural networks acquire knowledge for the detection, classification and localization of the fault facing different network conditions. The neural networks were implemented using NeuralWorks software. In this approach the three-phase voltage and current pre and post-fault values were utilized as inputs, for training and test purposes. The Alternative Transients Program (ATP) software was used to generate data for the transmission line (440 kV) in a faulted condition, both for the purposes of training and tests. The results obtained showed that the global performance of the ANN architectures is highly satisfactory for fault detection, classification and localization purposes. Considering all the studied cases, the ANN outputs converged to the correct levels very rapidly after fault occurrence.
Electric power systems; fault detection; classification and localization; artificial neural networks (ANNs)