Fault location algotithm based on ann for single line-to-ground fault in MV distribution system

Abstract

This project was focused to detect and locate the single line-to-ground (SLG) fault which occurs on transmission lines in MV networks. SLG fault is the most common compared to other faults. To detect and locate single-phase ground faults on MV networks, the ANN based fault location algorithm was developed to achieve this objective. In beginning, the MV network was designed by using Electromagnetic Transient Program-Alternative Transient Program (EMTP-ATP) according to a real network (IEEE 34 feeder). After that the network was simulated by creating faults at different locations. In addition, an earth resistance were changed and all the data which was used, is recorded to develop the ANN based FL algorithm. The fault current signal was measured in the EMTP program. Also, were analyzed captured signals by using the Fast Fourier Transform (FFT) in MATLAB to calculate the highest first three domains. 150 fault locations were simulated on the MV network to generate fault currents. The data is gathered from simulated fault currents. The data was trained and tested in the neural network in MATLAB to develop the algorithm. The developed algorithm is able to estimate a location for SLG faults which were simulated on the MV network. From a comparison between all location faults which were simulated and all locations which were detected in the developed FL algorithm it was found that the total error between them was small and the error is less than 0.72 km. Then based on the developed algorithm, the Simulink circuit was generated and tested to detect a SLG fault which may occur in the MV network in a future.