500 kV GIS隔离开关触头温度监测及外壳温度传感器优化布置  

Temperature Monitoring on Disconnector Contact of 500 kV GIS and Optimal Placement of Temperature Sensors on Enclosure

作　　者：曾小松 罗菁 姚强 苗玉龙 曾福平[2] ZENG Xiaosong;LUO Jing;YAO Qiang;MIAO Yulong;ZENG Fuping(Mian County Electric Power Supply Branch Company,State Grid Shaanxi Electric Power Company,Shannxi Hanzhong 724200,China;School of Electrical and Automation Engineering,Wuhan University,Wuhan 430072,China;Jiangbei Electric Power Supply Branch Company,State Grid Shaanxi Electric Power Company,Chongqing 401147,China;Electric Power Research Institute of Chongqing Electric Power Company,Chongqing 401123,China)

机构地区：[1]国网陕西省电力公司勉县供电分公司,陕西汉中724200 [2]武汉大学电气与自动化学院,武汉430072 [3]国网重庆市电力公司江北供电分公司,重庆401147 [4]国网重庆市电力公司电力科学研究院,重庆401123

出　　处：《高压电器》2021年第10期111-119,共9页High Voltage Apparatus

基　　金：国家自然科学基金资助(51607127)。

摘　　要：GIS触头温度是反应触头热故障信息最直接的指标,有必要实时监测其温度。目前无法在触头上安装传感器来监测其温度,因此文中提出利用多点外壳温度和环境温度的监测值来反演计算GIS触头温度,并对布置在外壳上的多点传感器进行了优化。首先建立了500 kV隔离开关气室的温度—流体场有限元分析模型,计算出温度场分布,确定了外壳传感器安装位置。同时进行温升试验,将试验和仿真结果对比,验证仿真模型的合理性。接着利用支持向量回归机训练多点外壳温度、环境温度和触头温度的对应关系,并利用多点外壳温度和环境温度的实测来反演计算触头温度,比较计算和实测结果发现平均绝对监测误差为0.47%。最后,利用遗传算法对外壳温度传感器进行优化布置,传感器数量从25个减少到5个,优化后平均绝对监测误差为2.07%,节约了成本,提高了反演计算的速度。Temperature of contact of GIS is the most direct indicator reflecting thermal fault informationof the contact and is necessary to be monitored real time.At present,it is impossible to install the sensor on the contact to monitor the temperature.Therefore,it is proposed in this paper to use the monitoring value of temperature at multiple point of the enclosure and the ambient temperature for inversion calculation of temperature of the contact of GIS.Also,the multiple-point sensors arranged on the enclosure is optimized.First,a temperature-fluid field finite element analysis model of 500 kV disconnector gas compartment is set up,the distribution of temperature field is calculated and the installation position of the sensors on the enclosure is determined.At the same time,the temperature rise test is performed to compare the test and simulation result in order to verify the rationality of the simulation model.Then,the support vector regression is used to train the corresponding relationship between multiple-point temperature of the enclosure,the ambient temperature and the contact temperature.The actual measurement of the multiple-temperature on the enclosure and ambient temperature is used for conversion calculation of contact temperature.It is found by way of comparing the calculated and actually measured result that the average absolute monitoring error is 0.47%.Finally,the optimal placement of the temperature sensor on the enclosure is performed by using the generic algorithm.The number of sensors is reduced from 25 to 5 and the average absolute monitoring error after optimization is 2.07%,which saves the cost and improves the speed of conversion calculation.

关 键 词：GIS 温度监测 温度反演 传感器优化布局 

分 类 号：TM564.1[电气工程—电器] TM595