An Improved Method Using Artificial Neural Network for Fault Detection and Fault Pole Identification in Voltage Source Converter-Based High-Voltage Direct Current Transmission Lines

Abstract

In this work, a method has been proposed for voltage source converter (VSC)-based high-voltage direct current (HVDC) transmission lines to detect the fault and find the fault pole during earthed and unearthed faults. Pre-processed features from DC voltage signals of both the poles and neutral to ground current signals from the rectifier end are taken as input to the proposed method. In this work, back-propagation neural network has been used as a classifier to detect the fault and to find the faulty pole. Three modules are designed—BNN-D for fault detection, BNN-P for fault pole identification and BNN-G for ground identification. The advantage of proposed method is that it not only detects the fault in a VSC-HVDC transmission lines, but it also identifies the faulty pole. Another advantage of proposed method is it does not need any communication link as it uses one-end measurement and the reach setting is up to 99.9% of the line length. Results of the proposed method have better selectivity, reliability, robustness and accuracy.