Small Sample Bearing Fault Diagnosis Method Based on Multi-physics Coupled Digital Twin

Wen-yang ZHANG , Chao ZHANG , Xi-ming ZHANG , Xiao-feng ZHOU , Xin-yu GE , Ke-xi XU , Wen-tao ZHAO

Science Technology and Engineering ›› 2026, Vol. 26 ›› Issue (31) : 13472 -13481.

PDF
Science Technology and Engineering ›› 2026, Vol. 26 ›› Issue (31) : 13472-13481 DOI: 10.12404/j.issn.1671-1815.2507972
research-article
Small Sample Bearing Fault Diagnosis Method Based on Multi-physics Coupled Digital Twin
Author information +
History +
PDF

Abstract

To address the challenges of sample scarcity and performance degradation in traditional deep learning-based bearing fault diagnosis, a novel fault diagnosis framework based on a multi-physical-field-coupled digital twin was proposed. A high-fidelity digital twin model was established by incorporating the bearing's geometric configuration and thermo-solid coupling effects. The model enabled realistic simulation of operational behavior. Twin datasets under diverse fault modes were generated to alleviate the lack of real experimental samples. A multi-scale convolutional neural network (MSCNN) was designed to improve multi-scale feature representation. The network was trained with a limited number of real samples to enhance diagnostic performance under small-sample conditions. Experiments were conducted on a bearing test bench platform to validate the proposed method. The results show that high diagnostic accuracy is achieved under varying operating conditions. Good generalization and stability are obtained. The proposed method provides an effective solution for intelligent fault diagnosis in small-sample scenarios.

Graphical abstract

Keywords

digital twin / multi-physical field coupling / bearing fault diagnosis / small sample learning

Cite this article

Download citation ▾
Wen-yang ZHANG, Chao ZHANG, Xi-ming ZHANG, Xiao-feng ZHOU, Xin-yu GE, Ke-xi XU, Wen-tao ZHAO. Small Sample Bearing Fault Diagnosis Method Based on Multi-physics Coupled Digital Twin. Science Technology and Engineering, 2026, 26(31): 13472-13481 DOI:10.12404/j.issn.1671-1815.2507972

登录浏览全文

4963

注册一个新账户 忘记密码

References

Funding

PDF (10037KB)

0

Accesses

0

Citation

Detail

Sections
Recommended

/

〈 〉