基于核典型相关分析融合的XLPE电缆局部放电模式识别

提出使用核典型相关分析方法提取XLPE电缆接头局部放电信号PRPD图谱特征信息,并使用K最近邻分类算法实现不同绝缘缺陷模式的高准确率识别。利用YJV-26/35 k V型电缆及其附件设计了4种典型绝缘缺陷,使用脉冲电流检测获取局部放电样本信息,绘制了PRPD图谱并应用于样本数据,研究不同特征向量下的识别效果,在适合维数最终获得较高识别正确率。相对于传统电力设备模式识别方法,不但可以有效反映信号非线性特征,并可以将多种特征进行有效融合,消除冗余特征。

Amalgamated Pattern Recognition of Partial Discharge in XLPE Cables Based on Kernel Canonical Correlation Analysis

This paper proposes that the characteristic information of PRPD spectrum of XLPE cable joints may be extracted in the kernel canonical correlation analysis method,and highly accurate recognition of different insulation defect modes is realized through the K nearest neighbor classification algorithm.YJV-26 / 35 kV cable and its accessories are used to design four typical insulation defects.Pulse current detection is used to obtain partial discharge sample information,and PRPD spectrums are drawn.The proposed method is applied to sample data to study recognition effects under different characteristic vectors.A quite high correct identification rate is finally obtained in the suitable dimension.Compared with patter recognition method of traditional power equipment,the proposed approach can not only effectively reflect nonlinear characteristics of the signal,but also effectively amalgamate various features and eliminate redundant features.