基于小波与分形理论的电力设备局部放电类型识别杜伯学.pdf
文本预览下载声明
DOI:10.13335/j.1000-3673.pst.2006.13.014
30 卷 13 期 电 网 技 术 Vol. 30 No. 13
2006 年7 月 Power System Technology Jul. 2006
文章编号:1000-3673 (2006 )13-0076-05 中图分类号:TM83 文献标识码:A 学科代码:470 4037
基于小波与分形理论的电力设备局部放电类型识别
杜伯学,魏国忠
(天津大学电气与自动化工程学院,天津市 南开区 300072 )
Partial Discharge Classification based on Wavelet and Fractal Theory
DU Bo-xue ,WEI Guo-zhong
(School of Electrical Engineering and Automation,Tianji n University,Nankai District ,Tianji n 300072,China )
ABSTRACT: On the basis of wavelet analysis a three- 立在交流电压的基础上,并不适用于高压直流系
dimensional time-frequency pattern to characterize partial 统。此外,采用指纹法或者其他统计方法提取放电
discharge (PD) impulse signal is upbuilt which comprehensively 特征存在的问题是:特征量在很大程度上受电压值
shows three basic features of PD impulse signals: time [10-11]
的影响,进而影响最终的识别结果 。传统的局
component, frequency component and distribution of
部放电识别方法主要集中在对局放脉冲信号的时
discharging energy. From the upbuilt three-dimensional pattern
the discharge features are extracted with fractal theory, thus PD 域或者频域特征进行分析,如采用脉冲波形特征或
fractal dimensions are used as feature vectors. The pattern 傅里叶变换等,对于非平稳的局部放电信号而言,
recognition of the type of PD signal is conducted by means of 这显然是不够的。小波变换为同时分析信号的时域
BP neural network. The discharge experiments have been 和频域特征提供了有力的工具,通过小波变换,将
conducted to validate the proposed method with five types 原来局限于时域或频域的一维信号扩展到二维的
artificial discharge models, and experiment results show that
显示全部