摘要
研究了电力变压器有载分接开关的故障诊断问题。对变压器分接开关的故障特性及原因分析后,考虑到传统支持向量机在诊断过程中效率低下、精确度差等缺点,提出了一种改进粒子群(PSO)优化支持向量机(SVM)的故障诊断方法。首先,对粒子群算法的惯性权值和学习因子做了相应改进,克服了PSO算法后期迭代精度不高的缺点;然后,利用改进后的PSO算法优化支持向量机的主要参数;最后,仿真结果表明,改进的PSO-SVM算法的诊断精度和速度均高于传统诊断方法,更适合在变压器分接开关诊断中应用。
Research the problem of tap‐changer fault diagnosis of transformer .Analyses the fault characteristics and the reasons of tap‐changer of transformer .Considering the problems of low efficiency ,poor accuracy in the diagnosis method based on traditional support vector . The improved PSO‐SVM （support vector machine ） was proposed to diagnose the faults in the paper .Firstly ,in order to avoid precision in the late time of the basic PSO algorithm ,the speed weights and the acceleration coefficient were adjusted .Secondly ,Using the improved PSO algorithm to optimize the parameters of support vector machine .Finally ,the simulation results show that the diagnostic accuracy and speed of the improved PSO‐SVM method is higher than traditional diagnosis method , and it is more suitable for the application of the transformer tap changer .
出处
《电子测量技术》
2016年第11期190-194,共5页
Electronic Measurement Technology
基金
国家自然科学基金(61104079)
河南省产学研基金(132107000027)资助项目
关键词
变压器分接开关
故障诊断
粒子群算法
支持向量机
tap-changer of transformer
fault diagnosis
PSO algorithm
support vector machine
作者简介
王福忠，1961年出生，教授，博导，主要从事工业过程控制、电力电子方面研究。
石秀立，1988年出生，硕士研究生，主要从事故障诊断研究。
