Convolutional Neural Network with Attention Mechanism and Visual Vibration Signal Analysis for Bearing Fault...

Area of Science:

• Mechanical Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Bearing faults are common in machinery, leading to failures.
• Traditional fault diagnosis relies on expert experience and time-frequency analysis.
• Existing intelligent methods often lack interpretability.

Purpose of the Study:

• To develop an interpretable deep learning model for bearing fault diagnosis.
• To enhance feature extraction and classification accuracy for bearing faults.
• To improve the understanding of diagnostic model decision-making.

Main Methods:

• A Convolutional Neural Network with an attention mechanism (CBAM-CNN) was developed.
• The Convolutional Block Attention Module (CBAM) was integrated for enhanced feature extraction.
• Gradient-Weighted Class Activation Mapping (Grad-CAM) was used for model interpretability.

Main Results:

• The CBAM-CNN achieved 99.81% accuracy on the experimental dataset.
• The model demonstrated improved convergence speed compared to a Base-CNN.
• Attention weight analysis showed distinct focus patterns for different fault types.
• Interpretability experiments revealed a focus on frequency distribution over amplitude.

Conclusions:

• The CBAM-CNN offers a highly accurate and interpretable solution for bearing fault diagnosis.
• The attention mechanism effectively enhances feature extraction in the time-frequency domain.
• Grad-CAM visualization provides insights into the model's decision-making process, highlighting frequency-based analysis.