基于小波分析和Kohonen神经网络的滚动轴承故障分析  

Fault diagnosis based on wavelet analysis and self-organizing feature map of roller bearings

作　　者：张梅军[1] 石文磊[1] 赵亮[1] 袁海龙[1] 

机构地区：[1]解放军理工大学工程兵工程学院,江苏南京210007

出　　处：《解放军理工大学学报（自然科学版）》2011年第2期161-164,共4页Journal of PLA University of Science and Technology(Natural Science Edition)

摘　　要：为了对旋转机械中滚动轴承的运行状态进行故障监测和诊断,在对振动信号进行采集和处理的基础上,提出了小波变换与Kohonen神经网络(SOM)相结合的滚动轴承故障诊断新方法。运用该方法在滚动轴承实验台上进行实验,用小波分析提取振动信号的特征值后,应用SOM网络对数据进行分类得到各种故障类型的标准样本,通过故障样本与标准样本的对比与分析得出诊断结论。结果表明,该方法能够准确的识别和诊断出滚动轴承的运行状态和故障类型,适合滚动轴承故障诊断,具有一定的工程实用价值。To monitor and diagnose the faults of roller bearing in the rotating machinery,a new method was presented which combines wavelet analysis with SOM based on the collection and the disposal of the roller bearing vibration signals. Experiments were carried out on the roller bearings lab-table and the eigenvalue by wavelet analysis.The fault diagnosis result was obtained by contrasting and analyzing the fault and the standard stylebook.The result show that the method can identify and diagnose not only the running states but also the fault types exactly.Therefore the method suits for the fault diagnosis of roller bearing,and is of applied value to engineering application.

关 键 词：故障诊断 小波分析 SOM网络 滚动轴承 

分 类 号：E92[军事—军事装备学] TK428[兵器科学与技术—武器系统与运用工程]