SOM神经网络在金属氧化物避雷器老化在线诊断中的应用
|
摘要
针对金属氧化物避雷器(MOA)老化在线诊断难的问题,提出一种基于SOM神经网络的MOA老化在线诊断方法。首先利用热成像仪获取已知老化情况的MOA温度分布数据,然后将其代入SOM神经网络进行训练。当SOM网络达到训练精度后,再利用网络对未知老化情况的MOA进行诊断,从而实现MOA老化在线诊断。验证表明:利用50只MOA温度分布数据对SOM网络训练500步后,其诊断正确率达到98%。使用10只MOA温度分布数据对SOM网络进行可靠性验证,诊断正确率100%。表明该方法具有较高可靠性。
Due to the difficult problem of metal oxide arresters aging on-line diagnosis. Based on selforganizing feature map neural network,a new method has been put forward. First,the thermal imager is used to get the heat distribution data of MOA. Second,plug the data into the SOM neural network for training. After the SOM network reaches the training accuracy,the network is used to diagnose the MOA of unknown aging,thereby achieving online diagnosis of MOA aging. Test results show that After 500 steps of training the SOM network with 50 MOA temperature distribution data,the diagnostic accuracy rate reached 98%. The reliability of the SOM network was verified using 10 MOA temperature distribution data,and the diagnostic accuracy rate was 100%. The results show that the method has high reliability.
Due to the difficult problem of metal oxide arresters aging on-line diagnosis. Based on selforganizing feature map neural network,a new method has been put forward. First,the thermal imager is used to get the heat distribution data of MOA. Second,plug the data into the SOM neural network for training. After the SOM network reaches the training accuracy,the network is used to diagnose the MOA of unknown aging,thereby achieving online diagnosis of MOA aging. Test results show that After 500 steps of training the SOM network with 50 MOA temperature distribution data,the diagnostic accuracy rate reached 98%. The reliability of the SOM network was verified using 10 MOA temperature distribution data,and the diagnostic accuracy rate was 100%. The results show that the method has high reliability.
引文
[1]张搏宇,李光范,陈立栋,等.避雷器及其比例单元的散热特性研究[J].电瓷避雷器,2010(4):46-50.ZHANG Boyu,LI Guangfan,CHEN Lidong,et al.Study on the thermal dispersion ability of MOA and MOA section[J].Insulators and Surge Arresters,2010(4):46-50.
[2]丁国成,李伟,刘云鹏,等.外部环境因素对MOA在线监测影响的试验[J].高电压技术,2008,34(6):1283-1287.DING Guocheng,LI Wei,LIU Yunpeng,et al.Experimental study of influence of environmental conditions on on-line monitoring data of metal oxide surge arrester[J].High Voltage Engineering,2008,34(6):1283-1287.
[3]周中山,张云峰,陈璞阳,等.多级电涌保护器对负载有效性试验分析[J].电瓷避雷器,2014,137(4):49-54.ZHOU Zhongshan,ZHANG Yunfeng,CHEN Puyang,et al.Analysis of glowing and arcing discharge performance of switching surge Protector[J].Insulators and Surge Arresters,2014,137(4):49-54.
[4]葛猛,韩学坤,陶安培.金属氧化物避雷器阀片老化缺陷的诊断及原因分析[J].高压电器,2009,45(3):145-147.GE Meng,HAN Xuekun,TAO Anpei.Diagnosising and analysising the aged flaw of MOA valve piece[J].High Voltage Apparatus,2009,45(3):145-147.
[5]杨仲江,张枨,柴健,等.氧化锌压敏电阻老化过程中非线性系数变化的研究[J].电子元件与材料,2011,30(9):27-30.YANG Zhongjiang,ZHANG Cheng,CHAI Jian,et al.Researchon the varying of nonlinear coefficient during the degradation of Zn O varistor[J].Electronic Components&Materials,2011,30(9):27-30.
[6]陈达波.人工鱼群算法在金属氧化物避雷器在线监测中的应用[J].电瓷避雷器,2016(3):105-109.CHEN Dabo.On-line monitoring of Metal Oxide Arrester Using Artificial Fish Swarm Algorithm[J].Insulators and Surge Arresters,2016(3):105-109.
[7]张志鹏,杨仲江.MOA谐波阻性电流补偿算法的研究[J].高压电器,2013,49(4):49-53.ZHANG Zhipeng,YANG Zhongjiang.Study on the harmonic resistive current compensation algorithm of MOA[J].High Voltage Apparatus,2013,49(4):49-53.
[8]陈楠.基于各次谐波法的MOA阻性电流检测方法探讨[J].华中电力,2009,22(6):60-63.CHEN Nan.Discussion on resistive current detection of MOA based on harmonic analysis method[J].Central China Electric Power,2009,22(6):60-63.
[9]徐志钮,赵丽娟,丁傲,等.一种新的SPD阻性电流提取算法[J].电力自动化设备,2010,30(12):47-51.XU Zhiniu,ZHAO Lijuan,DING Ao,et al.Calculation of SPD resistive current[J].Electric Power Automation Equipment,2010,30(12):47-51.
[10]E.T.Wanderley Neto,E.G.Costa,M.J.A.Maia,Artificial neural networks used for Zn O arresters diagnosis[J].IEEETransactions on Power Delivery,1994,24(3):51-52.
[11]鄢栋云.一种基于SOM网络预测剩余油分布的方法[J].电脑知识与技术,2016,12(2):161-162.YAN Dongyun.A method of predicting the remaining oil based on SOM network[J].Computer Knowledge and Technology,2016,12(2):161-162.
[12]许雅婧,黄小庆,曹一家,等.基于SOM神经网络聚类的空调负荷聚合方法[J].电力系统及其自动化学报,2015,27(11):26-33.XU Yajing,HUANG Xiaoqing,CAO Yijia,et al.Aggregation of air conditioner load based on self-organizing feature map neural network[J].Proceedings of the CSU-EPSA,2015,27(11):26-33.
[2]丁国成,李伟,刘云鹏,等.外部环境因素对MOA在线监测影响的试验[J].高电压技术,2008,34(6):1283-1287.DING Guocheng,LI Wei,LIU Yunpeng,et al.Experimental study of influence of environmental conditions on on-line monitoring data of metal oxide surge arrester[J].High Voltage Engineering,2008,34(6):1283-1287.
[3]周中山,张云峰,陈璞阳,等.多级电涌保护器对负载有效性试验分析[J].电瓷避雷器,2014,137(4):49-54.ZHOU Zhongshan,ZHANG Yunfeng,CHEN Puyang,et al.Analysis of glowing and arcing discharge performance of switching surge Protector[J].Insulators and Surge Arresters,2014,137(4):49-54.
[4]葛猛,韩学坤,陶安培.金属氧化物避雷器阀片老化缺陷的诊断及原因分析[J].高压电器,2009,45(3):145-147.GE Meng,HAN Xuekun,TAO Anpei.Diagnosising and analysising the aged flaw of MOA valve piece[J].High Voltage Apparatus,2009,45(3):145-147.
[5]杨仲江,张枨,柴健,等.氧化锌压敏电阻老化过程中非线性系数变化的研究[J].电子元件与材料,2011,30(9):27-30.YANG Zhongjiang,ZHANG Cheng,CHAI Jian,et al.Researchon the varying of nonlinear coefficient during the degradation of Zn O varistor[J].Electronic Components&Materials,2011,30(9):27-30.
[6]陈达波.人工鱼群算法在金属氧化物避雷器在线监测中的应用[J].电瓷避雷器,2016(3):105-109.CHEN Dabo.On-line monitoring of Metal Oxide Arrester Using Artificial Fish Swarm Algorithm[J].Insulators and Surge Arresters,2016(3):105-109.
[7]张志鹏,杨仲江.MOA谐波阻性电流补偿算法的研究[J].高压电器,2013,49(4):49-53.ZHANG Zhipeng,YANG Zhongjiang.Study on the harmonic resistive current compensation algorithm of MOA[J].High Voltage Apparatus,2013,49(4):49-53.
[8]陈楠.基于各次谐波法的MOA阻性电流检测方法探讨[J].华中电力,2009,22(6):60-63.CHEN Nan.Discussion on resistive current detection of MOA based on harmonic analysis method[J].Central China Electric Power,2009,22(6):60-63.
[9]徐志钮,赵丽娟,丁傲,等.一种新的SPD阻性电流提取算法[J].电力自动化设备,2010,30(12):47-51.XU Zhiniu,ZHAO Lijuan,DING Ao,et al.Calculation of SPD resistive current[J].Electric Power Automation Equipment,2010,30(12):47-51.
[10]E.T.Wanderley Neto,E.G.Costa,M.J.A.Maia,Artificial neural networks used for Zn O arresters diagnosis[J].IEEETransactions on Power Delivery,1994,24(3):51-52.
[11]鄢栋云.一种基于SOM网络预测剩余油分布的方法[J].电脑知识与技术,2016,12(2):161-162.YAN Dongyun.A method of predicting the remaining oil based on SOM network[J].Computer Knowledge and Technology,2016,12(2):161-162.
[12]许雅婧,黄小庆,曹一家,等.基于SOM神经网络聚类的空调负荷聚合方法[J].电力系统及其自动化学报,2015,27(11):26-33.XU Yajing,HUANG Xiaoqing,CAO Yijia,et al.Aggregation of air conditioner load based on self-organizing feature map neural network[J].Proceedings of the CSU-EPSA,2015,27(11):26-33.