基于IPSO-ELM的GIS局部放电模式识别研究  

Research on GIS Partial Discharge Pattern Recognition based on IPSO-ELM

作　　者：王鑫 王前 田俊波 WANG Xin;WANG Qian;TIAN Junbo(China Energy Investment Corporation Beihai Power Generation Co.,Ltd.,Guangxi Beihai 536000,China;School of Electrical Engineering,Guangxi University,Guangxi Nanning 530000,China)

机构地区：[1]国能广投北海发电有限公司,广西北海536000 [2]广西大学电气工程学院,广西南宁530000

出　　处：《广西电力》2024年第4期30-36,43,共8页Guangxi Electric Power

基　　金：国能广投北海发电有限公司科技创新项目(202400881)。

摘　　要：气体绝缘组合电器(Gas Insulated Switchgear,GIS)中的局部放电(Partial Discharge,PD)现象会对设备的长期稳定运行构成严重威胁,因此,准确有效地识别局部放电故障对保障电力系统的安全性至关重要。本文提出了一种基于特高频法(Ultra High Frequency,UHF)和改进粒子群优化-极限学习机(Improved Particle Swarm Optimization-Extreme Learning Machine,IPSO-ELM)模型的GIS局部放电故障类型识别方法,通过引入动态惯性权重和自适应变异操作的改进粒子群算法来优化极限学习机模型的关键参数,从而增强模型的分类性能,实现了对GIS局部放电信号的高效分类。测试集结果表明,改进后的模型分类准确率高达98.75%,可以有效解决目前诊断方法的不足。The partial discharge(PD)phenomenon in gas insulated switchgear(GIS)poses a serious threat to the long-term stable operation of equipment.Therefore,accurately and effectively identifying partial discharge fault is crucial for ensuring the security of power system.a GIS partial discharge fault type recognition method is proposed in this paper based on the ultra high frequency(UHF)technique and an improved particle swarm optimization-extreme learning machine(IPSO-ELM)model.By introducing a dynamic inertia weight and adaptive mutation operation into the improved particle swarm optimization algorithm,the key parameters of the extreme learning machine model are optimized,which enhances the model's classification performance.The method achieves efficient classification of GIS partial discharge signals,and test set results show that the improved model achieves a classification accuracy of 98.75%,which effectively addresses the limitations of current diagnostic methods.

关 键 词：气体绝缘组合电器 局部放电 特高频 粒子群算法 极限学习机 模式识别 

分 类 号：TM855[电气工程—高电压与绝缘技术]