绝缘子污秽放电声发射的统计指纹分析  

Statistical Fingerprint Analysis for Contaminated Insulator Acoustic Emission Signals

作　　者：李红玲[1] 文习山[1] 

机构地区：[1]武汉大学电气工程学院,武汉430072

出　　处：《高电压技术》2010年第11期2705-2710,共6页High Voltage Engineering

摘　　要：在深入研究绝缘子污秽放电声发射信号的基础上,提出采用统计指纹参量来评定绝缘子外绝缘状态的新方法。提出将污秽绝缘子外绝缘状态划分为3个阶段,即安全、报警和危险,为发布污闪预警信息提供依据。绝缘子污秽放电的强度和声发射信号之间存在对应关系,根据绝缘子污秽放电声发射信号的放电幅值分布和频谱分布图提取了17个统计指纹参量,绘制了绝缘子污秽放电不同发展阶段的统计指纹图谱,显著地表征了不同放电阶段的差异。并从17个统计参量中提取了3个主成分特征量。最后设计了基于最小二乘支持向量机的污秽绝缘子外绝缘状态分类器。实验结果证明了该统计指纹分析方法的可行性。Based on deep research of the acoustic emission signals emitted from contaminated insulators,a new statistical fingerprint analysis method is introduced to assess the external insulation state of contaminated insulator.It is proposed that the external insulation state of contaminated insulator can be divided into three stages,namely,safe,alarming and dangerous stages,providing a base to disseminate early flashover warning information.It is revealed that a corresponding relationship exists between the strength of filthy discharge and acoustic emission signals.According to the distribution diagram of discharge voltages and frequency spectrum,17 fingerprint parameters of acoustic emission signals were extracted.Then the fingerprint maps of different discharge development stages were protracted,and the differences among different discharge stages were significantly expressed.In order to simplify the structural design of the subsequent classifier,3 principal components were extracted from the 17 fingerprint parameters.Moreover,the external insulation state of contaminated insulator was assessed using the least squares support machine（LS-SVM） combined with principal component analysis.Finally,the feasibility of the statistical fingerprint analysis method is proved by the experimental results.

关 键 词：污秽 绝缘子 声发射 放电指纹 最小二乘支持向量机 分类识别 

分 类 号：TM852[电气工程—高电压与绝缘技术]