engleski

Detection of high impedance faults in power transmission network with nonlinear loads using artificial neural networks

High impedance faults (HIF) represent one of the most difficult problems for fault detection in the power transmission networks. The main difficulty during the detection of HIFs is that a fault current is very low and therefore it is difficult to detect a fault through conventional protection devices such as distance or overcurrent relays. In case when transmission network contains a variety of connected nonlinear loads the detection of fault is even more challenging, because HIF currents and nonlinear load currents may have similar RMS values. Therefore, it is important to make a distinction between HIF current and rated current of nonlinear loads. This paper presents an approach for distinguishing HIFs from nonlinear load operation using artificial neural networks (ANNs). A part of the 110 kV transmission network is modelled in MATLAB Simulink to perform simulations of HIFs and nonlinear load operation. As input values ANNs use voltage and current waveforms in the frequency domain from both ends of the line. Pattern recognition neural network is used to make a distinction between HIF current and nonlinear load current in case of similar RMS values. Two cases of ANNs with different network sizes are compared considering the number of neurons in hidden layer. The proposed approach is successfully tested with actual measured current of single-phase electric locomotive with diode converters.

High impedance faults ; nonlinear loads ; artificial neural networks ; power transmission system ; transmission lines ; relay protection

nije evidentirano