摘要
滚动轴承作为列车的重要组成部分,其工况状态直接影响列车的安全性能,因此对滚动轴承进行故障诊断的意义非比寻常。当轴承发生故障时会导致振动信号中出现冲击响应成分,因而可以通过对冲击响应成分进行特征提取来诊断故障。但是背景噪声的影响不可忽视,本文在时频分析法的基础上,将BP神经网络与时频分析相结合,分别建立了小波包-BP、EMD-BP两种诊断模型。通过对仿真信号的处理验证,结果表明小波包-BP模型能够更有效地去除噪声,诊断故障类型。
The safety of the trains can be directly influenced by the status of the rolling bearings, which are essential components of the trains.As a result,it is quite important to diagnose the locomotive bearing fault. Localized defects in bearings lead to periodical impulsive vibration, so they can be diagnosed by extracting impulsive response components. However, the noise under the practical environment should not be ignored. Based on joint time-frequency analysis, this paper presents a combination of neural network with time- frequency analysis, establishes two models, wavelet packet back propagation （BP） network and empirical mode decomposition （EMD） BP network. From the case study of locomotive bearing fault signal processing, it can be concluded that wavelet packet-BP has a better performance in eliminating noise and diagnose the locomotive bearing fault.
出处
《机电一体化》
2017年第4期21-27,72,共8页
Mechatronics
基金
国家自然科学基金资助项目(51505311)
江苏省自然科学基金资助项目(BK20150339)
关键词
轴承
故障诊断
小波包
EMD
神经网络
bearing fault diagnosis wavelet packets EMD neural network
