基于电流-角位移与PSO-SVM的GIS隔离开关机械故障诊断OA
Mechanical Fault Diagnosis of GIS Based on Current-Angular Displacement and PSO-SVM
针对目前采用单一特征进行气体绝缘组合电器(gas insulation switchgear,GIS)隔离开关机械故障诊断准确率较低、误判率较高及多特征故障诊断设备较复杂等问题,提出一种基于电流与角位移相结合的 GIS 隔离开关机械故障诊断方法.首先,通过 GIS 隔离开关分闸过程中所经历的 4 个阶段将电流曲线与角位移曲线分为四部分,提取曲线各阶段极值与平均值作为特征值,通过近邻成分分析(neighborhood components analysis,NCA)进行特征选择,将权重较高的特征组成诊断数据库;其次,通过粒子群优化(particle swarm optimization,PSO)算法优化支持向量机(sup-port vector machine,SVM)关键参数进行故障诊断;最后,以 ZF6-252 型隔离开关为例,通过本实验室自主研发的GIS 隔离开关故障诊断系统采集其在正常运行状态下及端盖锈蚀、脱销状态下的电流、角位移数据.试验结果表明,该方法分类准确率达到 97.78%,可为 GIS 隔离开关机械故障诊断提供一定参考.
In view of the current problems such as the low accuracy rate and high misjudgment rate of mechanical fault diagnosis of gas insulated switchgear(GIS)disconnectors using a single feature,as well as the complexity of multi-feature fault diagnosis equipment,it proposes a mechanical fault diagnosis method for GIS disconnectors based on the combination of current and angular displacement.Firstly,it divides the current curve and angular displacement curve into four parts based on the four stages experienced during the o-pening process of the GIS disconnector.It extracts the extreme values and average values of each stage of the curve as characteristic values.It carries out feature selection through neighbor component analysis(NCA),and combines the features with higher weights in-to a diagnostic database.Secondly,it optimizes the key parameters of the support vector machine(SVM)through the particle swarm optimization(PSO)algorithm for fault diagnosis.Finally,taking the ZF6-252 type disconnector as an example,it collects the current and angular displacement data of it under normal operation and in the state of rusted and out-of-pin end covers through the GIS discon-nector fault diagnosis system independently developed by our laboratory.The test results show that the classification accuracy of this method can reach 97.78%,providing a certain reference for the mechanical fault diagnosis of GIS disconnectors.
彭少卿;荆澜涛;田瑞;李金阔;王亮
辽宁省送变电工程有限公司,辽宁 沈阳 110072沈阳工程学院电气工程学院,辽宁 沈阳 110136国网辽宁省电力有限公司超高压分公司,辽宁 沈阳 110003中国能源建设集团辽宁电力勘测设计院有限公司,辽宁 沈阳 110179沈阳工程学院电气工程学院,辽宁 沈阳 110136
信息技术与安全科学
气体绝缘组合电器电机电流角位移粒子群优化SVM
GISmotor currentangular displacementPSOSVM
《东北电力技术》 2026 (9)
44-50,62,8
辽宁省教育厅基本科研项目(LJKMZ20221707)
评论