Bayesian variational transformer : a generalizable model for rotating machinery fault diagnosis
Xiao, Yiming and Shao, Haidong and Wang, Jie and Yan, Shen and Liu, Bin (2024) Bayesian variational transformer : a generalizable model for rotating machinery fault diagnosis. Mechanical Systems and Signal Processing, 207. 110936. ISSN 0888-3270 (https://doi.org/10.1016/j.ymssp.2023.110936)
Preview |
Text.
Filename: Xiao-etal-MSSP-2023-Bayesian-variational-transformer.pdf
Accepted Author Manuscript License: Download (1MB)| Preview |
Abstract
Transformer has been widely applied in the research of rotating machinery fault diagnosis due to its ability to explore the internal correlation of vibration signals. However, challenges still exist despite the countless efforts. Generally, Transformer is more prone to overfitting than CNN on small-scale datasets. In practical engineering, collecting sufficient fault samples for training is difficult, resulting in poor generalization of Transformer. In addition, the measured signals are often accompanied with severe noise, further reducing the generalization performance of the model. Meanwhile, the collected signals often follow different distributions due to the changing operating conditions, which places higher demands on the generalizability of Transformer. This paper proposes a Bayesian variational Transformer (Bayesformer) to cope with the abovementioned problems. In Bayesformer, all the attention weights are treated as latent random variables, rather than determined values as the previous studies. This allows to train an ensemble of networks, instead of a single one, enhancing the generalizability of the model. Three experimental studies are conducted to illustrate the developed model and superior diagnostic performance is showed throughout the experiments.
ORCID iDs
Xiao, Yiming, Shao, Haidong, Wang, Jie, Yan, Shen and Liu, Bin ORCID: https://orcid.org/0000-0002-3946-8124;-
-
Item type: Article ID code: 87699 Dates: DateEvent15 January 2024Published17 November 2023Published Online8 November 2023AcceptedSubjects: Technology > Mechanical engineering and machinery
Social Sciences > Industries. Land use. Labor > Management. Industrial ManagementDepartment: Strathclyde Business School > Management Science Depositing user: Pure Administrator Date deposited: 20 Dec 2023 16:20 Last modified: 26 Nov 2024 20:23 URI: https://strathprints.strath.ac.uk/id/eprint/87699