基于形态滤波和EMD-AR谱的轴承故障特征提取

Extraction of Fault Features for Bearings Based on Morphological Filtering and EMD-AR Spectrum

将AR谱估计与EMD方法结合,应用组合形态滤波对故障信号进行降噪预处理,对预处理后的信号进行EMD分解,之后对各阶IMF做AR谱估计并集总平均,从而提取振动信号的故障特征频率。文中所述算法能够避免HHT方法中Hilbert变换所产生的难以解释的负频率,较准确地提取出滚动轴承振动信号的故障特征频率,从而为滚动轴承振动信号的检测与故障诊断研究提供参考意见。

Combining the AR spectrum estimation method with EMD, the morphological filtering was applied to denoising pre-process of the fault signal. Then, the signal was decomposed by using EMD. The AR spectrum of each order IMF was estimated. The average value of the AR spectrums was calculated. Finally, the fault characteristic frequency of the vibration signal was extracted. In this method, the negative frequency generated by Hilbert transform in HHT can be avoided, and the fault characteristic frequency of rolling bearings can be extracted. This work may provide a reference for vibration detection and fault diagnosis of rolling bearings.