Research on bearing fault detection using a filtering algorithm based on multi-scale morphology and an improved PSO...

Area of Science:

• Mechanical Engineering
• Signal Processing
• Vibration Analysis

Background:

• Mathematical morphology filtering is common for bearing fault detection but struggles with noise and transient features.
• Existing methods lack robustness in complex noisy environments.

Purpose of the Study:

• To develop an enhanced multi-scale mathematical morphology filtering architecture for improved bearing fault detection.
• To address limitations in noise sensitivity and transient feature extraction.

Main Methods:

• Proposed an enhanced filtering architecture using mathematical morphology principles.
• Utilized characteristic frequency intensity coefficient for operator selection.
• Developed a multi-scale operator optimized with particle swarm optimization and a tent chaotic sequence.
• Validated the strategy with simulated and experimental faulty bearing data.

Main Results:

• The proposed multi-scale filtering strategy effectively captures weak periodic transient components in noisy vibration signals.
• Achieved superior accuracy in detecting inner ring, outer ring, and rolling element faults.
• Demonstrated enhanced performance in complex noise environments compared to conventional methods.

Conclusions:

• The enhanced multi-scale mathematical morphology filtering strategy offers a robust solution for bearing fault detection.
• The method significantly improves accuracy and noise resilience in vibration signal analysis.
• This approach provides a valuable tool for condition monitoring in rotating machinery.