Scholarlink

<p>High-impedance faults (HIFs) in distribution networks are characterized by low fault current levels and irregular arcing behaviors, making them easily confused with normal switching disturbances and difficult to detect for traditional protection. This paper proposes a morphology-based HIF detection scheme that combines an improved pattern spectrum with distribution-aware clustering. The Hadamard-matrix-based gradient pattern spectrum (HGPS) is developed, enabling quantitative characterization of multiscale waveform fluctuations and simultaneous description of positive and negative pulse components. Based on HGPS, a corresponding entropy index (HGPSE) is formulated to quantify fluctuation intensity and determine disturbance onset in a windowed manner. To achieve accurate classification, the Wasserstein-distance-based K-means clustering (WDKC) is adopted to measure the differences in morphological distribution patterns reflected by HGPS. Comprehensive simulation studies have been conducted to evaluate the performance of the proposed scheme under various conditions, including different noise levels, sampling frequencies, and network structures. The results demonstrate that the proposed scheme can effectively extract key morphological features of HIF signals and achieve accurate, reliable, and robust fault detection in complex distribution network environments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

High-impedance fault detection in distribution networks based on morphological pattern spectrum and Wasserstein K-means clustering

  • 期刊论文

  • 发表日期：

    2026年8月3日

查看全文
  • Mengshi Li,
  • Zhi Ding,
  • Tianyao Ji,
  • Qinghua Wu

摘要

High-impedance faults (HIFs) in distribution networks are characterized by low fault current levels and irregular arcing behaviors, making them easily confused with normal switching disturbances and difficult to detect for traditional protection. This paper proposes a morphology-based HIF detection scheme that combines an improved pattern spectrum with distribution-aware clustering. The Hadamard-matrix-based gradient pattern spectrum (HGPS) is developed, enabling quantitative characterization of multiscale waveform fluctuations and simultaneous description of positive and negative pulse components. Based on HGPS, a corresponding entropy index (HGPSE) is formulated to quantify fluctuation intensity and determine disturbance onset in a windowed manner. To achieve accurate classification, the Wasserstein-distance-based K-means clustering (WDKC) is adopted to measure the differences in morphological distribution patterns reflected by HGPS. Comprehensive simulation studies have been conducted to evaluate the performance of the proposed scheme under various conditions, including different noise levels, sampling frequencies, and network structures. The results demonstrate that the proposed scheme can effectively extract key morphological features of HIF signals and achieve accurate, reliable, and robust fault detection in complex distribution network environments.