Artificial neural network technique for transmission line protection on Nigerian power system

This paper presents a unique and efficient artificial neural network (ANN) based fault detection, classification and location on part of the Nigerian 132kV transmission line. The objective is to evaluate the performance of ANN based relays connected at both ends of the lines using feed-forward non-linear supervised back propagation algorithm with Levenberg-marguardt network topology. Using the PSCAD/EMTP software, the faults from both ends of the transmission lines are generated and fed into that same line using two different 132kV voltage sources with several variations of fault inception angle, location and resistance. The faults currents are then extracted, processed and divided into training and testing data using MATLAB software. The results obtained from the simulations are validated using real-data extracted from microprocessor based relay connected to Aba-Umuahia 132kVtransmission line. The results demonstrate the ability of ANN to correctly identify, classify and localize an actual fault occurring on that transmission line with high accuracy.