交流电压下GIS中导电微粒的模式识别方法研究  

Research on Pattern Recognition of Conducting Particles in GIS under AC Voltage

作　　者：马飞越 张涛 牛勃 丁培 杨朝旭 荣海军 MA Feiyue;ZHANG Tao;NIU Bo;DING Pei;YANG Zhaoxu;RONG Haijun(Electric Power Research Institute,State Grid Ningxia Electric Power Co.,Ltd.,Yinchuan Ningxia 750002,China;State Key Laboratory for Strength and Vibration of Mechanical Structures,School of Aerospace,Xi'an Jiaotong University,Xi'an Shaanxi 710049,China)

机构地区：[1]国网宁夏电力有限公司电力科学研究院,宁夏银川750002 [2]西安交通大学航天航空学院机械结构强度与振动国家重点实验室,陕西西安710049

出　　处：《湖北电力》2019年第6期28-34,共7页Hubei Electric Power

基　　金：国家自然科学基金面上项目(项目编号:61976172);国网宁夏电力有限公司科技项目(项目编号:5229DK17000N)。

摘　　要：气体绝缘金属封闭开关设备(Gas Insulated Switchgear,GIS)内部自由导电微粒是影响GIS运行可靠性的重要因素。为了实现GIS内部导电微粒的在线识别,为GIS设备的局部放电严重程度和危险度评估提供理论依据,建立了一套GIS设备局部放电缺陷的试验平台。利用模式识别的方法,在试验过程中采集不同导电微粒的超声波信号,通过对信号的分析和处理,提取出特征,将所提取的不同种类的导电微粒的特征输入到极限学习机(Extreme Learning Machine,ELM)中,通过训练和测试,最终的模型识别正确率能够达到92.58%。该方法为今后GIS设备内部自由导电微粒的识别带来了新思路。The internal free conducting particles of gas insulated switchgear(GIS)are important factors affecting the reliability of GIS operation.In order to realize the online identification of conducting particles in GIS,and to provides a theoretical basis for the partial discharge severity and risk assessment of GIS equipment,this paper establishes a test platform for partial discharge defects of GIS equipment.And then the paper uses a pattern recognition method to acquire ultrasonic signals of different conducting particles during the test,and extracts features by analyzing and processing signals.Finally,the characteristics of the extracted different kinds of conducting particles are input into the Extreme Learning Machine,andthrough continuous training and testing,the final model recognition accuracy rate reaches 92.58%.This method brings new ideas for the identification of free conducting particles in GIS equipment in the future.

关 键 词：气体绝缘金属封闭开关设备 导电微粒 局部放电 模式识别 极限学习机 

分 类 号：TM595[电气工程—电器] TM855