Periodicity-enhanced sparse representation for rolling bearing incipient fault detection

Area of Science:

• Mechanical Engineering
• Signal Processing
• Condition Monitoring

Background:

• Incipient fault detection in rolling bearings is challenging due to weak fault signatures obscured by significant background noise.
• Effective early detection of bearing faults is crucial for preventing catastrophic failures and ensuring operational reliability.

Purpose of the Study:

• To develop a novel periodicity-enhanced sparse representation method for improved incipient fault detection in rolling bearings.
• To address the challenge of weak fault feature extraction in noisy environments.

Main Methods:

• Proposed periodicity-enhanced basis pursuit denoising (PBPD) with theoretical derivation.
• Defined fault proportion to quantify sparse signal fault severity.
• Designed a periodicity-decision criterion for optimal fault period filtering.
• Investigated and adopted maximal overlapping discrete wavelet packet transform (MODWPT) as the linear transformation.
• Developed adaptive strategies for key PBPD parameter selection.

Main Results:

• PBPD effectively enhances weak fault features by reducing background noise.
• The periodicity-decision criterion accurately identifies optimal fault periods.
• Adaptive parameter selection ensures robust performance across different conditions.
• Simulations and experimental results validate the method's efficacy.

Conclusions:

• The developed periodicity-enhanced sparse representation method significantly improves incipient fault detection in rolling bearings.
• PBPD offers a robust solution for extracting fault information from noisy signals.
• This approach contributes to enhanced condition monitoring and predictive maintenance strategies for rotating machinery.