基于粒子滤波状态估计的滚动轴承故障识别方法

Fault Diagnosis of Rolling Bearing Based on Particle Filter State Estimation

提出了一种基于粒子滤波状态估计的滚动轴承故障识别方法,该方法主要包括故障模型建立和故障识别两个步骤;在故障模型建立部分,首先依据滚动轴承不同故障状态下的振动信号,建立对应的自回归模型,作为故障模型;在故障识别部分,将正常状态下对应的模型,转化为状态空间模型,设计粒子滤波器,然后对不同的故障状态进行估计,提取其残差的相关特征,并结合模型参数特征应用BP神经网络识别算法进行故障识别;最后以美国凯斯西储大学的滚动轴承振动数据为例,验证了该方法的有效性。

A fault diagnosis method of rolling bearing based on particle filter state estimation is proposed. The method mainly includes two steps： fault model establishment and fault identification. In the fault model establishment part, the corresponding autoregressive model is established according to the vibration signals of the rolling bearings in different fault states, which is used as the fault models In the fault recognition part, the corresponding model in the normal state is transformed into the state space model, and the particle filter is designed. Then, the related features of the residuals are extracted by estimating the different fault states. Combined with the model parameters, the BP neural network identification algorithm is used to identify the faults. Finally, the validity of the method is verified by the vibration data of rolling bearing of Case Western Reserve University.