An Intelligent Fault Diagnosis Method for Bearings with Variable Rotating Speed Based on Pythagorean Spatial Pyramid...

Area of Science:

• Mechanical Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Deep learning methods show promise for rotating machinery fault diagnosis.
• Existing methods struggle with variable rotating speeds, leading to low accuracy or complex adjustments.
• Industrial machinery often operates under variable working conditions.

Purpose of the Study:

• To develop a robust fault diagnosis method for bearings under variable rotating speeds.
• To improve accuracy and simplify parameter adjustment in bearing fault diagnosis.
• To enable a single deep learning model for diverse operating speeds.

Main Methods:

• Utilizing continuous wavelet transform scalogram (CWTS) to decompose vibration signals.
• Implementing a Pythagorean spatial pyramid pooling convolutional neural network (PSPP-CNN).
• Integrating a PSPP layer at the front of the CNN for enhanced feature extraction.

Main Results:

• The proposed CWTS and PSPP-CNN method demonstrates superior diagnostic accuracy for variable speed bearings.
• The PSPP-CNN model effectively processes CWTSs of varying scales.
• A single trained PSPP-CNN model can diagnose faults across the full working speed range.

Conclusions:

• The CWTS and PSPP-CNN approach offers a significant advancement in rotating machinery fault diagnosis.
• This method overcomes limitations of existing techniques for variable speed conditions.
• The developed model provides a generalized and accurate solution for bearing fault detection.