Early Fault Detection of Rolling Bearings Based on Time-Varying Filtering Empirical Mode Decomposition and Adaptive...

Area of Science:

• Mechanical Engineering
• Vibration Analysis
• Signal Processing

Background:

• Early fault detection in rolling bearings is crucial for preventing catastrophic failures.
• Strong background noise often hinders the effectiveness of traditional methods like time-varying filtering empirical mode decomposition (TVFEMD).
• Existing methods struggle to extract subtle fault characteristics in noisy conditions.

Purpose of the Study:

• To propose a new, robust method for early fault detection in rolling bearings.
• To enhance the extraction of fault characteristics from signals corrupted by strong background noise.
• To improve the reliability and accuracy of bearing fault diagnosis.

Main Methods:

• A novel weighted envelope spectrum kurtosis index is developed for selecting effective intrinsic mode functions (IMFs) from TVFEMD decomposition.
• A synthetic impact index (SII) is used with a gray wolf optimization algorithm to optimize multipoint optimal minimum entropy deconvolution adjusted (MOMEDA) parameters (fault period T, filter length L).
• The optimized MOMEDA method adaptively filters the reconstructed signal, followed by envelope spectrum analysis for fault feature extraction.

Main Results:

• The proposed method effectively extracts early fault features from simulated and measured rolling bearing signals.
• The filtered signals exhibit significantly higher first-order correlated kurtosis (FCK) and fault feature coefficient (FFC) compared to classical methods.
• The proposed method achieves lower sample entropy (SE) and envelope spectrum entropy (ESE), indicating superior noise suppression and feature enhancement.

Conclusions:

• The combined TVFEMD and optimized MOMEDA approach provides a reliable and effective solution for early fault detection in rolling bearings.
• This method demonstrates superior performance in noisy environments compared to FSK, MCKD, and the basic TVFEMD-MOMEDA techniques.
• The developed indices and optimization strategy enhance the sensitivity and accuracy of bearing fault diagnosis.