基于时变自回归模型与支持向量机的旋转机械故障诊断方法

旋转机械如电机振动信号普遍存在非平稳性，同时对于某台设备已有的故障样本往往非常有限的。针对这些问题，提出一种基于时变参数自回归模型和支持向量机相结合的旋转机械故障诊断方法。首先通过对振动信号建立TVAR模型，提取模型系数及残差的方差作为信号特征，然后利用SVM对信号进行分类，继而实现故障自动识别。转子实验台实验结果表明该文方法能够有效提取非平稳信号的特征，并能在较少训练样本的情况下获得较好的诊断结果。

TVAR and SVM Based Fault Diagnosis Method for Rotation Machine

Considering rotation machines like a motor,the vibration signals are of universal nonstationarity and the fault samples are usually very limited.Aiming at these problems,a new fault diagnosis method based on TVAR(time-varying autoregressive model) and SVM(support vectors machine) is presented.Firstly,the features including model parameters and the variance of remnant are extracted by modeling the vibration signals as a TVAR model.Secondly,a SVM is used to identify the fault pattern in terms of those features.Finally,experiments on a rotation machine test rig show that,this method can effectively extract features from nonstationary signals,and can get an excellent result despite of less training samples.