Artificial neural networks for single phase fault detection in resonant grounded power distribution systems

This paper presents a new protection algorithm for the detection of single phase-to-ground faults in a power distribution network with compensated neutral grounding. This algorithm, based on the comparison of the residual current and phase currents, uses artificial neural networks. This method is tested with EMTP simulations as well as with experimental data which are closer to reality. Hence, the efficiency of the algorithm is shown.