作者：王辉东，陈锋（国网浙江杭州市余杭区供电有限公司，浙江 杭州 310007）

摘要：高压开关柜发生局部放电时产生的超声波信号中存在着大量的信息，局部放电作为开关柜绝缘故障的重要征兆及表现方式，其类型的识别对于开关柜绝缘状态的评估具有重要的意义。为了准确地识别高压开关柜局部放电类型，采用经验模态分解（EMD）的方法对局放信号进行分解并提取能量信息，利用支持向量机（SVM）建立高压开关柜局部放电信号分类模型。实验结果验证了上述方法的有效性。为了解决SVM核函数g和非负惩罚因子C主观选取问题，运用灰狼算法（GWO）优化这两个参数。研究结果表明，与SVM、PSO-SVM和GA-SVM相比，GWOSVM可有效提高开关柜局放信号分类精度。

关键词：经验模态分解；灰狼算法；支持向量机；分类识别；遗传算法；粒子群算法

Abstract: There is a lot of information in ultrasonic signals generated when partial discharge occurs in high voltage switchgear. Partial discharge is an important sign and manifestation of insulation failure of switchgear. The identification of its type is of great significance for the assessment of insulation state of switchgear. In order to identify the partial discharge type of high voltage switchgear accurately, the empirical mode decomposition (EMD) method is used to decompose the local discharge signal and extract the energy information. A support vector machine (SVM) is used to establish the classification model of partial discharge signal of high voltage switchgear.Experimental results verify the effectiveness of the above methods. In order to solve the problem of subjective selection of SVM kernel function g and non-negative penalty factor C,the gray Wolf algorithm (GWO) was used to optimize these two parameters. Compared with SVM, PSO-SVM and GA-SVM,GWO-SVM can effectively improve the classification accuracy of switching cabinet signals.

Key words: Empirical modal decomposition; Gray wolf algorithm;Support vector machine; Classification and identification;Genetic algorithms; Particle swarm optimization

在线预览：基于EMD分解和GWO-SVM的开关柜局放信号识别

摘自《自动化博览》2019年12月刊