LAUSR

Abstract Short-circuit fault detection and classification in high-voltage direct-current (HVDC) electric power transmission lines are necessary for rapid location and removal of faults, as well as for recovering all or part of the power transmission capacity. In this study, a new and efficient technique is designed for protecting current-source converter-based HVDC (CSC-HVDC) lines. In this proposed method, new features considering the moving average and maximum coordinate difference criteria are extracted from local voltage and current signals measured with a relatively low sampling rate at the rectifier side. These extracted features provide excellent recognition to distinguish the external and internal short-circuit faults. The multiclass support vector machine model is also used to detect and classify different short-circuit faults in real-time operation. The comprehensive tests on a CSC-HVDC system verify the suggested protection strategy's high accuracy and dependability even under the circumstances not considered in the initial preparing and training stage. These results also authenticate the designed scheme's stability against external faults and lightning strikes, low sensitivity to measurement noises, and excellent performance in detecting and classifying high-resistance internal faults.