Improving rolling bearing online fault diagnostic performance based on multi-dimensional characteristics

Area of Science:

• Mechanical Engineering
• Signal Processing
• Condition Monitoring

Background:

• Rolling bearing failures are a major cause of rotating machinery damage and economic loss.
• Vibration signals from bearings are complex, nonlinear, and non-stationary, challenging traditional diagnostic methods.
• Accurate health assessment requires advanced techniques beyond standard time or frequency domain analysis.

Purpose of the Study:

• To develop a novel, multi-dimensional fault diagnostic method for rolling bearings.
• To achieve accurate diagnosis of various fault types and severities with real-time performance.
• To improve upon existing pattern recognition methods for bearing health monitoring.

Main Methods:

• Feature extraction using entropy, Holder coefficient, and improved generalized fractal box-counting dimensions from vibration signals.
• Intelligent bearing fault pattern recognition via a grey relation algorithm applied to extracted feature vectors.
• Integration of improved fractal dimension into multi-dimensional characteristics for enhanced diagnosis.

Main Results:

• The proposed method effectively extracts health status feature vectors.
• The grey relation algorithm successfully performs intelligent fault pattern recognition.
• Experimental validation demonstrates effective recognition of different fault types and severities.

Conclusions:

• The novel multi-dimensional characteristic-based method offers effective rolling bearing fault diagnosis.
• The integration of improved fractal box-counting dimension significantly enhances diagnostic accuracy.
• This approach provides a robust solution for real-time monitoring and fault identification in rotating machinery.