A Multi-component Bearing Fault Diagnosis Using Fast Iterative Filtering Technique

Condition monitoring (CM) signals of rolling element bearings in a rotating machine are typically non-linear, non-stationary signals which often exhibit multi-component amplitude modulation (AM)-frequency modulation (FM) characteristics. This poses a great challenge in signal analysis for an accurate detection of the faulty component(s) of a bearing in practical applications. Fast Iterative Filtering(FIF)is an effective technique for the analysis of multi-component and low signal-to-noise (SNR) signals. FIF uses certain iterating filters such as Toeplitz filters to quickly decompose a multi-component signal into intrinsic mode functions (IMFs) by means of fast Fourier transform. The technique is utilized in this study to decompose a multi-component bearing defect signal containing characteristic frequency components from inner and outer race faults as well as a faulty roller element. The bearing defect frequencies are then extracted from the most relevant IMF using envelope analysis. The result presented in this study validates that the proposed technique can detect the defect components in a CM signal while suppressing the mode mixing problem typically found in empirical mode decomposition (EMD) analysis. The comparison study presented in this paper shows that the proposed technique is more effective in the analysis of a multi-component bearing defect signal than the EMD algorithm.