Distortion-Based Detection of High Impedance Fault in Distribution Systems

Abstract

Detection of the high impedance fault (HIF) in distribution systems is significant for power utilization safety. In addition to the low fault currents, traditional approaches are invalid to detect HIFs also due to the diverse characteristics, including the slight HIF nonlinearity during the weak arcing process, the distortion offset caused by the lag of heat dissipation, and the interference of background noise. This paper proposes a distortion-based algorithm to improve the reliability of HIF detection under various conditions. Firstly, the challenges brought by the diversity of HIF distortions are explained according to the field experiments in a 10 kV real-world distribution system. HIFs are classified into five types according to the distortions of their current waveforms. Secondly, a definition of interval slope is introduced to describe waveform distortions. The interval slope is extracted by combining methods of linear least square filtering (LLSF) and Grubbs-criterion-based robust local regression smoothing (Grubbs-RLRS), so that the distortions under different fault conditions can be uniformly described. Thirdly, an algorithm is proposed to judge the features presented by the interval slope, and distinguish from non-fault conditions. Finally, the reliability and security of the proposed algorithm are thoroughly analyzed with real-world HIFs and the simulated HIFs obtained in IEEE 34-bus and IEEE 123-bus systems. Results show the improvements of the proposed algorithm by the comparisons with other advanced algorithms.