摘要
针对电动机轴承早期故障信号非线性非平稳性特征,造成故障信号特征提取和故障诊断困难,提出一种改进的基于添加自适应白噪声的完备集合经验模态分解与支持向量机结合的电动机轴承故障诊断方法。将美国凯斯西储大学测得的电动机轴承正常运行、滚动针体故障、外圈故障、内圈故障共4种信号分别用CEEMDAN和EEMD进行分解,得到多个模式分量,再将IMF能量法计算得到的特征向量引入支持向量机,进行电动机轴承故障识别。试验对比研究表明,该方法能更有效进行电动机轴承早期故障识别。
Because the fault signal of the motor bearing has nonlinear and non-stationary characteristics, it is difficult to extract the fault signal feature and make the fault diagnosis. This paper puts forward a complete set of empirical mode decomposition based on adaptive add white noise improvement and support vector machine with its fault diagnosis method. The four types of signals of the motor bearing normal operation, rolling needle fault, outer and inner race faults measured by Case Western Reserve University are decomposed by CEEMDAN and EEMD to get multiple mode component, and then the feature vector is calculated by IMF energy method, which is introduced to the support vector machine for the fault diagnosis. Experimental result shows that this method can be used to effectively make the incipient fault diagnosis for the motor bearing.
作者
卓仁雄
肖金凤
ZHUO Renxiong;XIAO Jinfeng(School of Electrical Engineering,University of South China,Hengyang 421000,China)
出处
《机械制造与自动化》
2019年第1期36-39,共4页
Machine Building & Automation
基金
湖南省教育厅重点项目(17A182)
关键词
电动机轴承
模态分解
固有模态函数
支持向量机
故障诊断
motor bearing
mode decomposition
intrinsic mode function
support vector machine
fault diagnosis
