基于FIF-CYCBD的滚动轴承故障特征提取方法研究-《郑州大学学报(工学版)》

文章信息/Info

Title:: Fault Feature Extraction Method of Rolling Bearings Based on FIF-CYCBD

Author(s):: LIU Y; LI L J; WANG Y; et al.; School of Mechanical and Power Engineering, Zhengzhou University;

Keywords:: Fasting filtering and decomposition (FIF); the maximum second -order cycle is stable and blindly solved (CYCBD); rolling bearing; feature extraction; cycle frequency

摘要:: 针对滚动轴承所处工况复杂、提取故障特征困难的问题,提出了一种基于快速迭代滤波分解( FIF) 和最大二阶循环平稳盲解卷积( CYCBD) 的故障特征提取方法。首先,通过利用 FIF 方法对源信号进行自适应分解,得到一系列本征模态分量;其次,依据相关系数准则对和源信号相关系数大于 0. 6 的分量进行重构,并根据 FIF 得到的分解结果设置合适的循环频率采集器;最后,利用 CYCBD 方法对重构后的信号进行解混去噪,对处理后的信号进行包络解调分析。仿真实验以及相关实验数据表明,所提方法具有良好的信噪分离效果,相较于信号中突出的噪声分量,处理得到的故障特征频率幅值高于噪声幅值,可以有效实现轴承故障频率及其倍频特征的提取。

Abstract:: Because of the complexity of working conditions, it is difficult to extract fault features from vibration signal of the rolling bears. In order to address this problem, a fault feature extraction method based on fast iterative filter decomposition ( FIF) and maximum second-order cyclostationarity blind deconvolution ( CY- CBD) method was proposed. Firstly, the fault signal of the rolling bearing was decomposed by FIF to obtain a series of intrinsic mode function. The components with the correlation coefficient of the source signal greater than 0. 6 were reconstructed, and the appropriate cycle frequency was set according to the decomposition result obtained by FIF. Then the CYCBD method was used to unmix and denoise the reconstructed signal. Finally, the processed signal was envelope demodulated to successfully extract the fault features. Compared with the prominent noise component in the signal, the amplitude of fault characteristic frequency obtained by processing was higher than that of noise. Therefore, the method proposed in this paper could effectively realize the extraction of bearing fault frequency and its frequency doubling characteristics.