采用独立成分分析的局部放电脉冲双通道提取技术研究

Pulse Extraction for Partial Discharge with Double Channels with Independent Component Analysis

简单介绍了独立成分分析(ICA)基本模型的假设、含混性问题及模型估计的等变化性.根据局部放电脉冲序列幅值的超高斯分布特性以及实际工况下的其他条件,得出了将ICA基本模型用于强背景白噪声下微弱局部放电脉冲信号提取的可行性.针对IEC60270规定的标准脉冲电流法,基于双通道超宽带检测,对试品放电状态下的脉冲提取进行了仿真,对直流下油纸绝缘针板模型进行局部放电实验,并给出直流下局部放电脉冲提取的ICA含混性消除方法.实验结果表明,该方法能够提取出被白噪声完全淹没的局部放电脉冲信号,可以恢复出局部放电波形之间相对幅值关系和单个脉冲波形以及脉冲极值所对应的时间点等重要局部放电信息.

The assumption, ambiguities, and solution of basic independent component analysis （ICA） model are briefly introduced. According to the features of white noise （gaussian） and partial discharge （PD） signals （supergaussian） on the probability distribution of magnitude, and the IEC 60220 standard PD measurement method, it indicates that the basic ICA model is feasible to suppress white noise of PD pulse signals. The simulation and on field tests for PD pulses extraction using ultra-wideband （UWB） detection technique with double channels are carried out, respectively. And the ambiguities elimination method for ICA model to extract PD pulses under HVDC is also presented. The results show that PD pulses are able to be exactly extracted from the strong white noise with keeping PD pulses amplitude of the relative relationship, the maximum amplitude corresponding to the time point, and the other important information of PD pulses.