基于宽带检测的局放脉冲波形快速特征提取技术

根据电工设备绝缘存在多局放的工况,提出研制基于单个脉冲的局放宽带检测与模式识别在线监测系统.并指出研制该系统需要解决脉冲群快速分类这一关键技术,其由脉冲波形的快速特征提取以及基于波形特征的快速聚类分析组成.本文针对100Ms/S采样率获取的宽带脉冲波形、-时间序列,提出分别使用幅值参数法、等效时频法和时频熵法对局放脉冲波形的特征参数进行提取.在对人工设置的多局放源产生的脉冲群数据进行波形特征提取后,使用模糊聚类对脉冲群波形特征提取结果进行了对比分析.结果表明:三种波形特征提取方法均能在一定程度上很好地分离干扰源而提取出局放数据.GIS母线尖刺缺陷的试验数据处理同样证实了该技术的有效性和可行性.这为研制基于单个脉冲的宽带局放检测与模式识别在线监测系统提供了实用的脉冲波形快速特征提取技术.

Fast feature extraction technique for PD pulse shape based on wideband detection

According to multi-PDs phenomenon in insulation of high voltage electrical equipments,a novel PD detection and pattern recognition system developed by wide-band technology based on a single pulse shape is proposed in this paper. To develop this system,the fast grouping technique for pulse sequence should be used,which is composed of fast feature extraction technology and clustering analysis method for PD pulse shape.Focus is made on Amplitude-Parameters,Equivalent Time-Frequency and Time-Frequency Entropy methods,which are used to make feature extraction for PD pulse shape-time sequence detected by wide-band technology with a sampling rate of 100Ms/S.After feature extraction made for noised multi-PD data,the results are compared by using fuzzy clustering method.The results show that those methods do a good separation for PD pulses with the same characteristics to a certain extent,respectively.And the analysis result of noised PD date of a gas insulated switch(GIS) shows that the feature extraction methods are effective and feasible for multi-PD pulses sequence grouping in practice. This provides a good method to make feature extraction for pulse shapes to develop the multi-PDs detection and pattern recognition system.