Bearing Fault Feature Extraction Method Based on Enhanced Differential Product Weighted Morphological Filtering

Area of Science:

• Mechanical Engineering
• Signal Processing
• Condition Monitoring

Background:

• Bearing vibration signals often contain noise and harmonic interference, obscuring critical fault information.
• Accurate bearing fault feature extraction is crucial for predictive maintenance and machinery health.
• Existing methods struggle with noise suppression and automatic parameter selection.

Purpose of the Study:

• To propose a novel algorithm, enhanced differential product weighted morphological filtering (EDPWMF), for robust bearing fault feature extraction.
• To address limitations in parameter selection for morphological operators.
• To improve the accuracy and reliability of bearing fault diagnosis.

Main Methods:

• An enhanced differential product weighted morphological operator (EDPWO) was developed by integrating differential product and weighted operations into basic morphological operators.
• A fault feature ratio (FFR) index was introduced for automatic determination of the structuring element (SE) length and optimal weighting factors.
• The EDPWFM algorithm was applied to simulation and experimental bearing vibration signals.

Main Results:

• The proposed EDPWMF algorithm effectively extracts bearing fault features from signals contaminated with noise and interference.
• Comparative analysis demonstrated superior performance of EDPWMF over traditional methods like AVG, STH, and EMDF.
• Automatic parameter optimization using FFR enhanced the method's practical applicability.

Conclusions:

• EDPWMF provides a significant advancement in bearing fault feature extraction, particularly in noisy environments.
• The method offers a reliable and automated approach for fault diagnosis.
• This study contributes valuable insights for developing sophisticated morphological analysis techniques.