基于PSO和SVM的牵引变压器绝缘故障诊断  

Insulation fault diagnosis for traction transformer based on PSO and SVM

作　　者：方科[1] 苏雁军[1] 黄友文[1] 

机构地区：[1]暨南大学珠海学院电气自动化研究所,广东珠海519070

出　　处：《计算机辅助工程》2010年第3期83-86,共4页Computer Aided Engineering

基　　金：国家自然科学基金(50878188);铁道部科技研究开发计划(2008J002)

摘　　要：为提高牵引变压器绝缘故障诊断的正确性,在分析其负荷特征和特征气体产生机理的基础上,针对其故障特点提出基于粒子群优化(Particle Swarm Optimization,PSO)算法和支持向量机(Support Vector Machine,SVM)的牵引变压器绝缘故障诊断方法.根据罗杰斯比值法将变压器状态分为12种故障模式;用PSO算法优化SVM参数,充分发挥SVM具有较高泛化能力的优势.试验表明该方法能快速、准确地找到相应的优化参数,有效进行牵引变压器绝缘的故障诊断.To improve the accuracy of insulation fault diagnosis for traction transformer,according to its fault characteristics,an insulation fault diagnosis method based on Particle Swarm Optimization（PSO） algorithm and Support Vector Machine（SVM） is presented by analyzing the load characteristics of traction transformer and the generation mechanism of characteristic gases.The transformer states are classified into twelve kinds of fault patterns by using Rogers ratio method;PSO algorithm is applied to optimize SVM parameters,so as to bring SVM＇s advantage of high generalization ability into full play.The test indicates that the method can be used to determine the corresponding optimization parameters quickly and accurately,and thus the fault diagnosis for traction transformer can be carried out effectively.

关 键 词：牵引变压器 故障诊断 特征气体 粒子群优化 支持向量机 

分 类 号：TM922.73[电气工程—电力电子与电力传动]