基于天牛须搜索优化支持向量机的变压器故障诊断研究  

Research on transformer fault diagnosis based on a beetle antennae search optimized support vector machine

作　　者：方涛 钱晔 郭灿杰 宋闯 王志华 罗建平 巴全科 FANG Tao;QIAN Ye;GUO Canjie;SONG Chuang;WANG Zhihua;LUO Jianping;BA Quanke(Luoyang Power Supply Company,State Grid Henan Electric Power Company,Luoyang 471000,China;State Grid Henan Electric Power Research Institute,Zhengzhou 450052,China;Wuhan Kemov Electric Co.,Ltd.,Wuhan 430023,China)

机构地区：[1]国网河南省电力公司洛阳供电公司,河南洛阳471000 [2]国网河南省电力公司电力科学研究院,河南郑州450052 [3]武汉凯默电气有限公司,湖北武汉430023

出　　处：《电力系统保护与控制》2020年第20期90-96,共7页Power System Protection and Control

基　　金：国家电网公司总部科技项目资助(52170218000M);国网河南省电力公司2019年科技项目资助(5217A01801U5)。

摘　　要：为了准确地判断变压器绕组是否出现故障,保证变压器供电的可靠性,提出了一种基于天牛须搜索算法优化支持向量机(BAS-SVM)的变压器绕组故障诊断方法。采用支持向量机(SVM)作为变压器绕组形变程度的分类器,并应用天牛须算法对SVM的核函数和惩罚因子进行优化,通过人工经验训练BAS-SVM,使其具有很高的故障诊断精度。为了比较BAS-SVM算法在变压器绕组故障诊断的优越性,采用改进的粒子群优化算法(MPSO)优化SVM。通过仿真验证,BAS-SVM算法的故障诊断准确率比MPSO-SVM算法的故障诊断准确率高10%。最后通过实例验证了BAS-SVM算法对变压器绕组故障诊断的可行性。In order to accurately judge whether a transformer winding has faults and ensure the reliability of power supply of the transformer,a method of transformer winding fault diagnosis based on BAS-SVM is proposed.It uses an SVM as the classifier of the transformer winding deformation degree,and optimizes the kernel function and penalty factor of the SVM by using a beetle antennae search algorithm.The BAS-SVM is trained by artificial experience to ensure that the algorithm has a high accuracy of fault diagnosis.In order to compare the advantages of the BAS-SVM algorithm in this application,a Modified Particle Swarm Optimization(MPSO)is also used to optimize SVM.The simulation results show that the fault diagnosis accuracy rate of the BAS-SVM algorithm is 10%higher than that of MPSO-SVM algorithm.Finally,the effectiveness of the BAS-SVM method on transformer winding fault diagnosis is verified by an example.

关 键 词：变压器 故障诊断 BAS-SVM 绕组变形 MPSO-SVM 

分 类 号：TM41[电气工程—电器]