基于Volterra-PARAFAC模型的滚动轴承故障诊断方法

Fault diagnosis method for rolling bearing based on Volterra-PARAFAC model

为解决Volterra模型用于复杂机械系统非线性特征提取时存在估计参数过多的问题,提出了一种新的Volterra-PARAFAC预测模型.在非线性特征提取中,所提出的预测模型的估计参数数目大大低于传统的Volterra预测模型参数,有效地避免了维数灾难问题.在Volterra-PARAFAC预测模型辨识过程中,利用最小均方自适应(LMS)算法估计Volterra-PARAFAC预测模型的核参数向量,从而精确描述非线性系统.利用该方法对滚动轴承多种故障状态下的振动信号进行分析,得到的特征向量具有非常好的分类性能.试验结果表明,该方法能有效提取复杂机械系统的非线性特征,并能准确对不同状态下的滚动轴承故障信号进行分类.相比于传统的Volterra模型故障诊断方法,所提方法能够更准确地对滚动轴承故障进行诊断.

Estimating a huge number of parameters is the main drawback of Volterra prediction model applied in the nonlinear feature extraction of complex mechanical systems.To overcome this drawback,a new Volterra-PARAFAC (parallel factor analysis) prediction model is presented.In the nonlinear feature extraction,the number of estimated parameters of the proposed prediction model is much lower than that of the traditional Volterra prediction model,and the dimension curse can be avoided.effectively.In the identification process of Volterra-PARAFAC prediction model,the parameter vector of Volterra-PARAFAC prediction model is estimated by the least mean square method so as to accurately describe nonlinear systems.The vibration signals of several rolling bearing faults are analyzed by the proposed method,and the obtained parameter vectors have excellent classification performance.Experimental results show that this method can effectively extract the nonlinear characteristics of complex mechanical systems,and the failure recognition of rolling bearing in different working states can be realized accurately.Compared with traditional fault diagnosis methods based on the Volterra model,the proposed method can improve the precision of rolling bearing fault diagnosis.