Scholarlink

<p>Composite insulators are widely used in power systems. However, surface defects can compromise their insulation performance, with different defects having varying impacts. The effective identification of these defects and the development of maintenance strategies tailored to each type of defect are crucial. This paper presents a defect detection method that integrates electric field distribution with machine learning. First, a 3D model of the composite insulator is created using COMSOL Multiphysics software. Electric field simulations are conducted on the umbrella skirt surface of the insulator to obtain electric field distributions under different defect conditions. Feature engineering is then employed to extract relevant features from the electric field data, which are subsequently input into machine learning algorithms for classification. Experimental results demonstrate that this method effectively identifies defects on the composite insulator’s umbrella skirt surface, achieving defect identification accuracy of 92.50%. Furthermore, the machine learning classifier exhibits strong performance in terms of precision, recall, and F1 score. This approach significantly enhances the automation of defect recognition, providing valuable support for fault detection and maintenance decision-making in power systems.</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

Composite Insulator Skirt Defect Identification Based on the Electric Field Distribution Method Combined with Machine Learning

  • 期刊论文

  • 发表日期：

    2025年11月16日

查看全文
  • Shuaiqi Xu,
  • Changjin Che,
  • Qiang Yao,
  • Yunzhong Gao,
  • Zengyan Wang

摘要

Composite insulators are widely used in power systems. However, surface defects can compromise their insulation performance, with different defects having varying impacts. The effective identification of these defects and the development of maintenance strategies tailored to each type of defect are crucial. This paper presents a defect detection method that integrates electric field distribution with machine learning. First, a 3D model of the composite insulator is created using COMSOL Multiphysics software. Electric field simulations are conducted on the umbrella skirt surface of the insulator to obtain electric field distributions under different defect conditions. Feature engineering is then employed to extract relevant features from the electric field data, which are subsequently input into machine learning algorithms for classification. Experimental results demonstrate that this method effectively identifies defects on the composite insulator’s umbrella skirt surface, achieving defect identification accuracy of 92.50%. Furthermore, the machine learning classifier exhibits strong performance in terms of precision, recall, and F1 score. This approach significantly enhances the automation of defect recognition, providing valuable support for fault detection and maintenance decision-making in power systems.