摘要
维护电力系统中气体绝缘封闭组合开关设备的稳定运行,通过SF_6气体分解组分对设备绝缘状态进行早期的故障诊断是其中关键步骤。为了提升短期SF_6数据存在识别率不高技术难点,首先提出基于信息熵理论和主成分分析法的综合评价方法。然后建立数据变量特征评价模型,对于气体特征量选取有指导意义。其次通过搭建局部放电实验平台对SF_6气体进行分解实验,得到的短期数据进行方法验证。最后,使用原数据与通过验证后新特征变量数据进行多分类SVM分类比对,结果表明,满足评价模型数据可以提高其识别正确率。
To maintain stable operation of the gas-insulated closed combination switchgear in power system, early fault diagnosis of the equipment insulation state through the SF6 gas decomposition components is a vital step. To improve the recognition rate of short-term SF6 data, this study put forward a comprehensive evaluation method based on the information entropy theory and the principal component analysis method. And a data variable feature evaluation model was established for selection of characteristic variables of gas. Then, decomposition experiment of SF6 gas was carried out by setting up a partial discharge experimental platform to obtain the short-term data. Classification com- parison via multi-classification SVM between the original data and the verified new characteristic variable data shows that the recognition accuracy could be improved if the evaluation model data are satisfied.
作者
王博伟
杨同忠
王先培
赵乐
WANG Bowei;YANG Tongzhong;WANG Xianpei;ZHAO Le(School of Electrical Engineering,Wuhan University,Wuhan 430072,China;School of Electronic Information,Wuhan University,Wuhan 430072,China)
出处
《高压电器》
CAS
CSCD
北大核心
2018年第11期55-61,共7页
High Voltage Apparatus
基金
国家自然科学基金(50677047)~~
