基于多分类最小二乘支持向量机和改进粒子群优化算法的电力变压器故障诊断方法

发布时间：2018-01-04 15:18

本文关键词：基于多分类最小二乘支持向量机和改进粒子群优化算法的电力变压器故障诊断方法　出处：《高电压技术》2014年11期 　论文类型：期刊论文

【摘要】：为了提高故障诊断的准确率,提出了一种多分类最小二乘支持向量机(LS-SVM)和改进粒子群优化(PSO)相结合的电力变压器故障诊断方法。引入最小输出编码构造多个2分类LS-SVM,实现了变压器诊断的多类分类。利用PSO算法获得LS-SVM诊断模型的最优参数,并采用交叉验证原理来提高分类算法的整体泛化性能。实例分析结果表明,采用LS-SVM和PSO算法可以准确、有效地对变压器进行故障诊断;与传统的电力变压器故障诊断方法相比,该方法的诊断准确率更高。
[Abstract]:In order to improve the accuracy of fault diagnosis. A multi-classification least squares support vector machine (LS-SVM) and improved particle swarm optimization (PSO) are proposed. Combined power transformer fault diagnosis method. The minimum output code is introduced to construct multiple two-class LS-SVM. The multi-class classification of transformer diagnosis is realized, and the optimal parameters of LS-SVM diagnosis model are obtained by using PSO algorithm. The principle of cross validation is adopted to improve the generalization performance of the classification algorithm. The results of the example analysis show that the fault diagnosis of transformer can be done accurately and effectively by using LS-SVM and PSO algorithms. Compared with the traditional power transformer fault diagnosis method, the accuracy of this method is higher.
【作者单位】：国网河南省电力公司电力科学研究院;
【分类号】：TM41
【正文快照】： 0引言电力变压器是电网中进行能量转换、传输的核心,是电网安全运行的关键设备[1-2]。我国目前已有较多的变压器,其运行年限已超过15 a,设备自身存在着诸如绝缘老化等各种故障隐患[3],发生事故的概率也不断增加。一旦电力变压器发生事故,则可能会引起电网大停电,对中国社会经

【参考文献】