基于RBF神经网络的高压隔离开关过热状态预警方法  

Overheat State Assessment Method of High-voltage Disconnecting Switch Based on RBF Neural Network

作　　者：李勇 陈长杰 董玉林 

机构地区：[1]国网温州供电公司,温州市325000 [2]国网长沙供电公司,长沙市410015

出　　处：《自动化与仪器仪表》2016年第2期67-69,共3页Automation & Instrumentation

摘　　要：针对目前高压隔离开关的发热温度检测没有可靠在线监测设备和技术手段,导致难以对过热缺陷进行可靠预警的问题。本文基于RBF神经网络,综合考虑了影响高压隔离开关发热的负荷电流率、污秽等级和环境温度3个因素,建立了一个隔离开关过热故障的预警模型,通过对现场数据样本测试发现:模型对发热状态评估的总体正确率达到了94.44%,且做到了对过热缺陷100%的预警。Currently,there is no reliable on-line monitoring technique or equipment to monitor the temperature of the high-voltage disconnecting switch,and so it is difficult to detect the overheat faults and send out early warning signals effectively and immediately.Hence,an assessment model was built based on radical basis function( RBF) neural network. The model takes three factors into consideration. The three factors are the ratio of load current and rated current,pollution grade and ambient temperature which impact the temperature of the high-voltage disconnecting switch. To prove the validity of the assessment model,some field samples were used to test the model. The test result showed that the overall accuracy rate reached 94. 44%,and it can detect the overheat defects with 100%.

关 键 词：隔离开关 过热 RBF神经网络 状态预警 

分 类 号：TM564.1[电气工程—电器] TP183[自动化与计算机技术—控制理论与控制工程]