摘要
针对电磁斥力机构真空快速开关的机械状态监测问题,提出了一种最大奇异值能量熵(energy entropy of maximum singular value,EEMSE)和随机森林的故障诊断方法。首先,在真空快速开关中采集振动信号,对振动信号进行改进S变换得到模矩阵,随后对该矩阵的子矩阵进行奇异值分解,再利用信息熵理论对最大奇异值求熵得到特征向量,最后将特征向量输入随机森林模型进行故障分类和诊断。与不同特征量和分类器比较后的结果表明,文中提出的真空快速开关机械故障诊断方法特征一致性好,模型诊断速度较快,对实验样本总体诊断准确率达到了100%。
Aiming at the problem of mechanical state monitoring of vacuum fast switch based on electromagnetic repulsion mechanism,a kind of fault diagnosis method based on energy entropy of maximum singular value(EEMSE)and random forest is proposed.Firstly,the vibration signal is collected in the vacuum fast switch,and the vibration signal is transformed by improved S-transform to obtain the modular matrix.Then,the submatrix of the matrix is decomposed into singular value,and the feature vector is obtained by calculating the entropy of the maximum singular value by using the information entropy theory.Finally,the feature vector is input into the random forest model for fault classification and diagnosis.The result shows that after comparing with different feature and classifiers the mechanical fault diagnosis method of vacuum fast switch proposed in this paper has good feature consistency,fast model diagnosis speed and the overall diagnosis accuracy of experimental samples reaches 100%.
作者
张宇
毕凡
苏海博
王勇
刘俊翔
郑方晴
刁均伟
ZHANG Yu;BI Fan;SU Haibo;WANG Yong;LIU Junxiang;ZHENG Fangqing;DIAO Junwei(Test and Research Institute,Guangzhou Power Supply Bureau,Guangdong Power Grid Co.,Ltd.,Guangzhou 510410,China)
出处
《高压电器》
CAS
CSCD
北大核心
2023年第11期184-192,共9页
High Voltage Apparatus
基金
广东电网有限责任公司广州供电局科技项目(SF6罐式结构真空快速开关状态监测技术研究及应用)。
关键词
真空快速开关
改进S变换
奇异值分解
随机森林
故障诊断
vacuum fast switch
improved S-transform
singular value decomposition
random forest
fault diagnosis