局放模式识别中常用波形特征及其识别效率之比较  

Comparison of common waveform characteristic and recognition efficiency in PD pattern recognition

作　　者：曹昌睿 郑重[2] 

机构地区：[1]西宁供电公司,青海西宁810003 [2]华北电力大学,北京102206

出　　处：《青海大学学报（自然科学版）》2010年第6期41-44,48,共5页Journal of Qinghai University(Natural Science)

基　　金：国家重点基础研究发展计划973项目(2009CB724508)

摘　　要：文中研究了分段线性法、多项式拟合法、自回归模型法、物理特征法等4种不同的特征提取法,并对各种方法的识别效率进行了测定。结果表明,利用局部放电物理特征来进行局部放电模式识别的方法是最简单而有效的。Computer aided auto recognition of various types of partial discharges based on the pulse waveforms was investigated through application of an artificial neural network.Various feature extraction methods were applied,including the piecewise linearization,the polynomial fitting,the autoregressive model,and the physical characteristics of pulse waveforms.The results indicate that the method based on the physical characteristics of the partial discharge,which employs expert prior knowledge,is most effective and computationally least intensive.

关 键 词：局部放电模式 波形特征 物理特征 

分 类 号：TM401.1[电气工程—电器]