Evaluation of the Output Structures from Artificial Neural Networks on the Performance of Fault Detection in Power Distribution Systems

This paper presents and describes a method based on artificial neural networks for single-phase faults detection in power distribution systems. It was evaluated the impact from the number of neurons in the output layer structure and the number of hidden neurons from the networks. The neural networks data inputs are current phasors measured from each phase provided by the allocated PMUs on the distribution system. The testing system used for the simulations is the IEEE 34 Bus. The proposed methodology aims to identify in which phase the single-phase fault occurs ranging from 0 to 20 ohms the fault resistance.