New Method for Diagnosing Faults in Rolling Bearings Using ConvNeXt Network
Author Information
Author(s): Song Jiahao, Nie Xiaobo, Wu Chuang, Zheng Naiwei
Primary Institution: Inner Mongolia University of Technology
Hypothesis
Can a new fault diagnosis method combining ConvNeXt and improved DenseBlock effectively diagnose rolling bearing faults in noisy environments and with insufficient sample sizes?
Conclusion
The proposed method demonstrates high diagnostic accuracy and robustness in noisy environments and with limited training samples.
Supporting Evidence
- The method achieved the highest accuracy in diagnostic experiments across various working conditions and noise environments.
- Comparative experiments showed strong fault diagnosis capability even with noise pollution and insufficient training samples.
- The introduction of the DY-ReLU function significantly improved feature extraction and diagnostic accuracy.
Takeaway
This study created a smart way to find problems in machine parts called rolling bearings, even when there's a lot of noise or not enough data.
Methodology
The method uses continuous wavelet transform to process vibration signals and combines ConvNeXt with an improved DenseBlock for feature extraction.
Limitations
The method requires manual labeling of samples and has high computational resource demands.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website