基于EMD-SVD的高压电缆局部放电信号降噪算法研究  

Noise reduction method of power cable partial discharge signal based on EMD-SVD

作　　者：张小牛[1] 景俊伟 林健[1] 张哲 马星河[3] ZHANG Xiaoniu;JING Junwei;LIN Jian;ZHANG Zhe;MA Xinghe(Pingdingshan Tianan Coal Co.,Ltd.,Pingdingshan 467000,China;Henan Antenna and Power Company Limited,Pingdingshan 467000,China;Henan Polytechnic University,Jiaozuo 454000,China)

机构地区：[1]平顶山天安煤业股份有限公司,河南平顶山467000 [2]河南天通电力有限公司,河南平顶山467000 [3]河南理工大学,河南焦作454000

出　　处：《电子设计工程》2023年第22期96-100,共5页Electronic Design Engineering

基　　金：国家自然科学基金资助项目(U1404522)。

摘　　要：局部放电法是目前高压电缆在线绝缘监测常用的手段,高压电缆局部放电信号(PD)常常被多种噪音信号污染,其中以白噪音最为频繁。采用经验模态分解(Empirical Mode Decomposition,EMD)联合奇异值分解(Singular Value Decomposition,SVD)对PD信号进行降噪处理。PD信号被噪声污染后对其进行分解并得到若干个本征模态函数和残余分量,通过排列熵筛选出特征信号与含噪特征信号,用含噪的模态分量IMF构建Hankel矩阵,分解矩阵的奇异值,利用奇异值筛选完成对PD信号的去噪。实验表明,该方法降噪能力强,并具备自适应的特点。Partial Discharge(PD)is a common method for on-line insulation monitoring of high-voltage cables.PD signals of high-voltage cables are often polluted by various noise signals,among which white noise is the most frequent.(Empirical Mode Decomposition,EMD)and(Singular Value Decomposition,SVD)are used to denoise PD signals.The PD signal is decomposed after being polluted by noise,and several intrinsic modal functions and residual components are obtained.The characteristic signals and noisy characteristic signals are screened out by permutation entropy.The noisy modal component IMF is built into Hankel matrix,the singular values of the matrix are decomposed,and the PD signal is denoised by singular value screening.Experiments show that this method has strong noise reduction ability and adaptive characteristics.

关 键 词：局部放电 经验模态分解 奇异值 降噪 

分 类 号：TM85[电气工程—高电压与绝缘技术]