电子器件2018，Vol.41Issue(3)：644-648,5.DOI:10.3969/j.issn.1005-9490.2018.03.019
基于声信号分析的开关柜局部放电检测算法研究
Research on Partial Discharge Detection Algorithm for Switchgear Based on Acoustic Signal Analysis
唐云辉 1胡曦琳1
作者信息
- 1. 国电南瑞科技股份有限公司,南京211106
- 折叠
摘要
Abstract
Fault detection of high voltage switchgear is one of the factors that affect the safety of power grid. The ear-ly ultrasonic detection method has many problems,such as high cost and low detection efficiency. For this reason,an algorithm for partial discharge detection in switchgear based on spectrogram features is proposed. Firstly,the spec-trogram of the signal is calculated,and then the second order normalized central moments are calculated. Based on the calculated features,the deep learning network is constructed and enhanced based on the parameter sparsity. The experiments of partial discharge detection in switchgear show that the proposed features can improve the efficiency of fault detection. Compared to the algorithm based on Mel Frequency Ceptral Coefficient,the proposed algorithm im-proves the recognition rate of 2.5％. In addition,combined with the deep learning algorithm,the recognition rate can reach 99.8％.关键词
局部放电/语谱图/深度学习网络/美尔倒谱特征Key words
partial discharge/spectrogram/deep learning network/Mel Frequency Ceptral Coefficient分类
信息技术与安全科学引用本文复制引用
唐云辉,胡曦琳..基于声信号分析的开关柜局部放电检测算法研究[J].电子器件,2018,41(3):644-648,5.
