基于LSTM⁃MIV神经网络的SF_(6)断路器触头电寿命预测  

Electrical Life Prediction for Contact of SF_(6) Circuit Breaker Based on LSTM⁃MIV Neural Network

作　　者：马飞越 李澳 吴诚威 于家英 陈磊 倪辉 MA Feiyue;LI Ao;WU Chengwei;YU Jiaying;CHEN Lei;NI Hui(State Grid Ningxia Electric Power Research Institute,Yinchuan 750001,China;School of Electrical Engineering,Xi’an Jiaotong University,Xi’an 710049,China)

机构地区：[1]国网宁夏电力有限公司电力科学研究院,银川750001 [2]西安交通大学电气工程学院,西安710049

出　　处：《高压电器》2024年第2期69-77,共9页High Voltage Apparatus

基　　金：国网宁夏电力有限公司科技项目资助(5229DK20055)。

摘　　要：文中基于LW30⁃40.5型SF_(6)断路器全寿命烧蚀试验数据,采用归一化异常指数F反映断路器触头剩余电寿命,0表示电寿命完备,1表示电寿命耗尽。根据断路器分闸动态电阻—行程曲线,提出了将动态电阻—行程曲线上行程大于8 mm且电阻值大于500μΩ的第一个峰值点作为断路器主触头接触阶段与弧触头接触阶段分界点。据此得到反映断路器触头异常指数F的5个特征参数,采用皮尔逊相关系数法对特征参数进行独立性分析。通过LSTM神经网络建立了基于特征参数与异常指数F之间的回归预测模型,F回归预测均方差和绝对平均误差分别为0.014921和0.013053,采用MIV方法确定了各特征参数对异常指数F影响权重。Based on the lifetime ablation test data of LW30⁃40.5 SF_(6) circuit breaker,normalized anomaly index F is adopted in this paper to reflect the remaining electrical life of contact of the circuit breaker,where 0 indicates com⁃plete electrical life and 1 indicates exhaustion of electrical life.According to the dynamic resistance⁃stroke curve of the circuit breaker,the first peak point with a stroke greater than 8 mm and a resistance greater than 500μΩon the dynamic resistance⁃stroke curve is proposed as the dividing point between the contacting stage of main contact of and the contacting stage of arc contact of the circuit breaker.Based on this,five characteristic parameters reflecting the ab⁃normal index F of contact of the circuit breaker are obtained.The Pearson correlation coefficient method is used to an⁃alyze the characteristic parameters independently.LSTM neural network is used to set up a regression prediction mod⁃el on the basis of relationship between the characteristic parameters and the anomaly index F.The mean square er⁃ror and absolute mean error of the regression prediction of the F index are 0.014921 and 0.013053,respectively.The MIV method is used to determine the influence weights of each characteristic parameter on the anomaly index F.

关 键 词：SF_(6)断路器 电寿命 动态电阻 LSTM⁃MIV神经网络 回归预测 

分 类 号：TM561[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]