Bearing Weak Fault Diagnosis Based on EMD-Wavelet Packet Denoising and Fast Spectrum Kurtosis

With the advent of Industry 4.0, the manufacturing industry is undergoing a profound transformation, where fault diagnosis of large rotating equipment in smart factories has become an important research area. In actual operation, rotating equipment, especially rolling bearings, serve as critical components that bear significant loads and are prone to compound faults. To address the difficulty in extracting fault features from vibration signals of rolling bearings, which poses challenges for fault diagnosis, this paper proposes a fault diagnosis method combining Empirical Mode Decomposition (EMD) and wavelet packet denoising with a fast spectral kurtosis algorithm. Specifically, the collected vibration signals are first decomposed using EMD, and effective signal components are selected based on weighted kurtosis values to reconstruct the signal. Subsequently, the selected signal components are denoised using a wavelet packet denoising algorithm. Then, the fast spectral kurtosis algorithm is applied to the reconstructed signal to extract the most significant fault information frequency bands and perform filtering. Finally, the Teager demodulation and 1.5-dimensional spectrum analysis are combined to extract bearing fault features, enabling effective fault diagnosis. Simulation and experimental data results show that the proposed method can effectively extract fault features of rolling bearings, achieving high diagnostic accuracy and practicality.