Fault feature extraction of rolling bearing using Birge-Massart threshold denoising with EEMD and spectral kurtosis

For the difficulty of extracting fault feature by adopting the traditional resonance demodulation method due to noise interference, an effective fault feature extraction method of rolling bearing was proposed by integrating Birge-Massart-based threshold de-noising strategy with ensemble empirical mode decomposition (EEMD) and fast spectral kurtosis algorithm. EEMD was applied to decompose original fault signals and then extract intrinsic mode function (IMF) components containing fault information by kurtosis criterion. Birge-Massart strategy and fast spectral kurtosis were employed to filter and denoise fault signals. The fault features of the filtered signals were extracted by Hilbert envelope demodulation. Through the fault feature extraction of the simulated signals and experimental signals based on this developed method, the results demonstrate that the proposed method is effective to improve the signal-to-noise ratio of fault signals and to acquire the frequency feature of inner and outer ring faults. Using kurtosis factor criterion to select IMF components can effectively retain the impact features of the original fault signal and remove the influence of irrelevant IMF components.