Bearing Fault-Detection Method Based on Improved Grey Wolf Algorithm to Optimize Parameters of Multistable Stochastic...

Area of Science:

• Mechanical Engineering
• Signal Processing
• Artificial Intelligence

Background:

• Traditional stochastic resonance systems lack adaptive parameter adjustment for bearing fault detection.
• Effective bearing fault diagnosis is crucial for industrial equipment reliability.

Purpose of the Study:

• To propose an adaptive-parameter bearing fault-detection method.
• To enhance the optimization performance for improved fault signal detection.

Main Methods:

• Improved grey wolf optimization algorithm incorporating Sobol sequence initialization, exponential convergence factor, adaptive position update, and Cauchy-Gaussian hybrid variation.
• Optimization of multistable stochastic resonance model parameters using the enhanced algorithm.
• Validation on open-source bearing datasets and a single-crystal furnace lifting device bearing.

Main Results:

• Accurate diagnosis of bearing faults with identified fault frequencies (e.g., 158 Hz inner ring, 162 Hz outer ring, 35 Hz lifting device).
• Experimental results align with theoretical predictions.
• Demonstrated faster convergence speed and higher output signal-to-noise ratio compared to other improved algorithms.

Conclusions:

• The proposed adaptive-parameter method effectively detects bearing faults.
• The enhanced grey wolf optimization algorithm significantly improves diagnostic performance.
• This method provides a reliable solution for bearing fault diagnosis in industrial applications.