Rolling Bearing Fault Diagnosis Using a Slope-Weighted, Limited Penetrable Horizontal Visibility Graph and Related...

Area of Science:

• Mechanical Engineering
• Signal Processing
• Data Analysis

Background:

• Rolling bearing fault diagnosis is crucial for mechanical system reliability and safety.
• Traditional graph signal methods using WLPHVG show promise but have limitations in capturing subtle dynamic changes.
• Existing methods struggle with selecting optimal parameters and enhancing spectral features for accurate health status monitoring.

Purpose of the Study:

• To develop an improved graph-based method for more accurate rolling bearing fault diagnosis.
• To enhance the sensitivity of graph spectral features to vibration signal dynamics.
• To introduce a novel fault feature indicator for efficient and effective fault detection.

Main Methods:

• Developed a slope-weighted, limited penetrable horizontal visibility graph (SWLPHVG) by optimizing visibility distance and employing a slope-weighting approach.
• Enhanced the responsiveness of the graph spectrum to vibration signal characteristics.
• Proposed a graph spectral area indicator for fault diagnosis, analyzing amplitude distribution and clustering.

Main Results:

• The SWLPHVG method significantly improved the detection of dynamic changes in vibration signals compared to traditional WLPHVG.
• The proposed graph spectral area indicator effectively captured fault characteristics while enhancing processing efficiency.
• Experimental validation confirmed the superiority of the SWLPHVG approach for rolling bearing fault diagnosis.

Conclusions:

• The SWLPHVG method offers a more sensitive and efficient approach to rolling bearing fault diagnosis.
• The developed graph spectral area indicator provides a robust tool for analyzing spectral features.
• This research contributes to advancing the reliability and safety of mechanical systems through improved fault detection techniques.