基于傅里叶分解与奇异值差分谱的滚动轴承故障诊断方法

针对强噪声条件下滚动轴承故障冲击特征难以提取的特点,提出了一种基于傅里叶分解与奇异值差分谱的滚动轴承故障诊断方法。首先通过傅里叶分解将非平稳的原始轴承故障振动信号分解为若干个固有频带函数,然后运用互相关系数法筛选固有频带函数进行信号重构,并对重构后的信号进行奇异值差分谱降噪,最后对联合降噪后的信号进行Hilbert包络谱分析,准确地识别出故障特征频率,进行故障诊断。仿真分析和试验都很好地验证了该方法的有效性。

Rolling Bearing Fault Diagnosis Based on FDM and Singular Value Difference Spectrum

Based on the characteristic that is the feature extraction of rolling bearing’s impact features is very hard under strong noise, a method based on Fourier decomposition method (FDM) and singular value difference spectrum is proposed. First, the non-stationary original bearing fault vibration signal was decomposed into several Fourier intrinsic band functions (FIBFs) by FDM. Then, the original signal was reconstructed by using correlation cross-coefficient method. The reconstructed signal was de-noised by the singular value difference spectrum. Finally, the fault characteristic frequency is accurately identified by using Hilbert envelope spectrum to the combined de-noised signal. The simulation analysis and test are good to verify the proposed method.