电器与能效管理技术 ›› 2020, Vol. 0 ›› Issue (3): 38-42.doi: 10.16628/j.cnki.2095-8188.2020.03.006

• 研究与分析 • 上一篇    下一篇

基于频带能量特征提取的10 kV固体绝缘柜表面放电音频信号识别*

侯春光, 孙小涵, 赵长权, 曹云东   

  1. 沈阳工业大学电器新技术与应用研究所, 辽宁 沈阳 110870
  • 收稿日期:2019-10-10 出版日期:2020-03-30 发布日期:2020-04-21
  • 作者简介:侯春光(1978—),男,博士,副教授,主要从事智能电器、电器设备智能化、泛在配电物联网等方面的研究工作。孙小涵(1995—),女,硕士研究生,主要从事数字信号处理、绝缘状态监测研究工作。赵长权(1994—),男,硕士研究生,主要从事电器设备状态监测研究工作。
  • 基金资助:
    * 国家自然科学基金项目(51977132)

Surface Discharge Audio Signal Recognition of 10 kV Solid Insulation Cabinet Based on Band Energy Feature Extraction

HOU Chunguang, SUN Xiaohan, ZHAO Changquan, CAO Yundong   

  1. Institute of Electrical New Technology and Application,Shenyang University of Technology, Shenyang 110870, China
  • Received:2019-10-10 Online:2020-03-30 Published:2020-04-21

摘要: 为准确监测10 kV固体柜的绝缘状况,提出了一种基于小波包频带能量特征识别的绝缘状态识别的方法。使用音频捕获设备获取固体绝缘柜表面放电及工作环境的音频信号;建立小波包基,通过正交分解得到高频信息和低频信息;通过分析音频信号的频谱特征来进行适当的分割,计算每个频率段的能量;最后,将每个频率段的能量归一化来建立特征向量,作为状态识别的依据。将表面放电声音和环境噪声的特征向量输入到支持向量机中,进行绝缘状态的分类和识别。通过试验数据分析,验证了小波包分析和支持向量机结合可以准确地确定固体绝缘开关设备的表面放电。

关键词: 沿面放电, 音频信号, 小波包分析, 支持向量机

Abstract: In order to accurately monitor the insulation condition of 10 kV solid cabinets,a method based on wavelet packet frequency band energy identification for insulation state identification was proposed.An audio capture device is used to obtain audio signals from the surface discharge of the solid cabinet and the working environment.A wavelet packet base is established,and high frequency information and low frequency information are obtained by orthogonal decomposition.The energy of each frequency segment is calculated by analyzing the spectral characteristics of the audio signal for proper segmentation.Finally,the energy of each frequency segment is normalized to establish a feature vector as the basis for state recognition.The feature vectors of surface discharge sound and ambient noise are input into a support vector machine to classify and identify the insulation state.Through the analysis of experimental data,it is verified that the combination of wavelet packet analysis and support vector machine can accurately determine the surface discharge of solid insulated switchgear.

Key words: surface discharge, audio signal, wavelet packet analysis, support vector machine

中图分类号: