滚动轴承故障的振动信号诊断方法

A hybrid diagnosis method based on vibration signals for rolling bearing fault

针对滚动轴承损伤类故障振动信号的特点,充分利用HMM、SVM在序列行为的分类和小样本方面的优势,把SVM的输出转化为HMM中观察值概率矩阵模型,建立了动态过程时间序列分类器,提高模型的学习速度和分类性;基于对包络解调信号提取AR模型参数构建的用于训练和故障识别的特征矢量,提出了一种基于SVM-HMM混合算法的滚动轴承故障诊断方法。将该方法应用到滚动轴承故障诊断中取得了较好的效果。

The vibration signals of faulty rolling bearings are dynamic and nonlinear,this makes fault identification become very difficult. In consideration of the classification ability of SVM and the distinguish ability of HMM to the dynamic time series,based on the characteristic vectors that are built up by extracting the AR model parameters from the envelope demodulation signal,it proposes a new method of rolling bearing fault diagnosis. Experiments show the effectiveness of the method.