改进LMD和LS-SVM在小电流接地故障选线中的应用  

Improved Local Mean Decomposition and Least Squares Support Vector Machines Applied in Fault Selection for Small Current Grounding System

作　　者：曹丽芳[1] 赵朋程[1] 陈颖[1] 王玉田[1] 张淑清[1] 张航飞[1] 徐剑涛[1] CAO Li-fang ZHAO Peng-cheng CHEN Ying WANG Yu-tian ZHANG Shu-qing ZHANG Hang-fei XU Jian-tao(Institute of Electrical Engineering, the Key Lab of Measurement Technology and Instrumentation of Hebei Province, Yanshan University, Qinhuangdao, Hebei 066004, Chin)

机构地区：[1]燕山大学电气工程学院河北省测试计量技术及仪器重点实验室

出　　处：《计量学报》2016年第6期632-637,共6页Acta Metrologica Sinica

基　　金：国家自然科学基金(61471312,61077071);河北省自然科学基金(F2015203413)

摘　　要：提出一种改进的局部均值分解（LMD）和最小二乘支持向量机（LS-SVM）相结合的小电流接地故障选线新方法。针对LMD存在端点效应的缺陷，提出了一种最小平方距离相关的改进算法，对端点效应进行了有效的抑制；LS-SVM在SVM基础上，用二次损失函数代替不敏感损失函数，用等式约束代替不等式约束，降低了计算复杂度。与径向基神经网络（RBF）方法的分类效果对比，验证了LS-SVM在非线性模式识别方面的优势。实验表明该方法能够很好地选出故障线路，为小电流接地故障选线提供了一种有效的新方法。A new fault line selection method for the small current grounding system based on improved local mean decomposition（ LMD ） combined with the least squares support vector machines （LS-SVM） is put forward. Against defects in end effect of LMD, the minimum squared distance-related to improve algorithm is proposed, and end effect is effectively suppressed ; LS-SVM is the developed algorithm on the basis of SVM, with a quadratic loss function instead of insensitive loss function, equality constraints instead of inequality constraints, reducing the computational complexity greatly. Compared with classification results of radial basis function neural network（RBF） , least squares support vector machine has advantages in nonlinear pattern recognition. The experiment showed that this method could be well selected fault line, provides an effective new method for fault section of small grounding lines.

关 键 词：计量学 故障选线 小电流接地 局部均值分解 端点效应 LMD改进算法 最小二乘支持向量机 

分 类 号：TB971[一般工业技术—计量学]