Scholarlink

Abstract <p>Partial discharge (PD) detection is a common technique for assessing the state of insulation in high voltage equipment. To suppress noise in PD signals of electrical equipment, a denoising method combining with singular value decomposition (SVD) and generalized S-transform (GST) is proposed. The Hankel matrix based on the signal time series is constructed and the SVD is carried out to eliminate the periodic narrowband noise, whose singular values have the feature of appearing in pairs. According to the time-frequency diagram of GST, the PD signal with white noise was then intercepted. Finally, the effective sequence number K of the singular value was determined according to the standard deviation change of the normalized singular value subset to suppress the white noise. To evaluate the denoising effect, the algorithm was compared with two other algorithms. The simulation and measurement results show that this work could effectively suppress noise interference of PD with advantages of good noise suppression and high feature preservation.</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

A Partial Discharge Noise Suppression Method Based on Singular Value Decomposition and Generalized S-transform

  • 期刊论文

  • 发表日期：

    2025年10月1日

查看全文
  • Zhanquan Wang,
  • Junhong Xing,
  • Linna Wang,
  • Donghui Zhang,
  • Yun Liu,
  • Mingxing Jiao,
  • Weimin Xia

摘要

Abstract

Partial discharge (PD) detection is a common technique for assessing the state of insulation in high voltage equipment. To suppress noise in PD signals of electrical equipment, a denoising method combining with singular value decomposition (SVD) and generalized S-transform (GST) is proposed. The Hankel matrix based on the signal time series is constructed and the SVD is carried out to eliminate the periodic narrowband noise, whose singular values have the feature of appearing in pairs. According to the time-frequency diagram of GST, the PD signal with white noise was then intercepted. Finally, the effective sequence number K of the singular value was determined according to the standard deviation change of the normalized singular value subset to suppress the white noise. To evaluate the denoising effect, the algorithm was compared with two other algorithms. The simulation and measurement results show that this work could effectively suppress noise interference of PD with advantages of good noise suppression and high feature preservation.