AutoVMDPgram: An Effective Method for Fault Diagnosis of Rolling Bearing

Area of Science:

• Mechanical Engineering
• Signal Processing
• Condition Monitoring

Background:

• The VMDPgram, combining VMD and WPT, shows promise in bearing fault diagnosis.
• Existing methods require further refinement for optimal performance in rotating machinery diagnostics.

Purpose of the Study:

• To address limitations in the VMDPgram for bearing fault diagnosis.
• To develop an enhanced method for detecting localized faults in rotating machinery.

Main Methods:

• Introduced unbiased autocorrelation (AC) function to analyze the cyclostationarity of vibration signals.
• Developed AutoVMDPgram by calculating kurtosis of unbiased AC of the squared envelope of sub-intrinsic modal functions (sub-IMFs).
• Incorporated modified adaptive resonance bandwidth (MARB) and a cumulative evaluation index for optimal sub-IMF selection.

Main Results:

• AutoVMDPgram effectively identifies bearing faults using simulated and experimental data.
• The method demonstrates superior performance compared to Autogram, spectral kurtosis (SKs), and VMD.
• The proposed cumulative index accurately locates optimal sub-IMFs without prior resonant frequency range determination.

Conclusions:

• AutoVMDPgram offers a robust and improved approach for bearing fault diagnosis.
• The integration of unbiased AC and advanced signal processing techniques enhances fault detection capabilities.
• This method provides a valuable tool for condition monitoring of rotating machinery.