Integrated fuzzy subdomain and test-time adaptation framework for cross-condition bearing RUL prediction
Measurement Science and Technology · 08.05.2026 · 37 (19) · 195105
DOI: 10.1088/1361-6501/ae65bf ↗Kurz gefasst
Fachartikel in der Zeitschrift Measurement Science and Technology, online erschienen am 12.05.2026. Verfasst von Zimeng Fan und 6 weiteren Autor:innen. Die Arbeit nennt 42 Literaturangaben. Gefördert von: Key Innovation Project of the Technology and Engineering Center for Space Utilization.
Abstract (Originalsprache)
Cross-condition bearing remaining useful life (RUL) prediction is frequently complicated by distribution shifts and the progressive, multi-stage nature of degradation. Conventional unsupervised domain adaptation methods often perform global alignment over the entire trajectory, which can obscure stage-dependent dynamics and cause negative transfer when mismatched stages are forced to align. To address this issue, we propose an integrated fuzzy subdomain and test-time adaptation framework for robust RUL prediction under unseen operating conditions. Specifically, we employ fuzzy stage modeling to enable stage-consistent cross-domain alignment while optimizing the source-domain regression objective. To further reduce residual deployment-time shifts, we incorporate constrained pseudo-label retraining at test time to construct prior-consistent pseudo RUL trajectories and calibrate the predictor with minimal parameter updates. Experiments on a public bearing benchmark across multiple transfer settings demonstrate that the proposed method improves robustness and accuracy over representative deep prognostics baselines.
Förderung
- Key Innovation Project of the Technology and Engineering Center for Space Utilization
Zitieren
APA
Fan, Z., Du, J., Jiao, D., Gui, X., Ning, S., Song, L., & Li, X. (2026). Integrated fuzzy subdomain and test-time adaptation framework for cross-condition bearing RUL prediction. Measurement Science and Technology, 37(19), 195105. https://doi.org/10.1088/1361-6501/ae65bf
Vancouver
Fan Z, Du J, Jiao D, Gui X, Ning S, Song L, et al. Integrated fuzzy subdomain and test-time adaptation framework for cross-condition bearing RUL prediction. Measurement Science and Technology. 2026;37(19):195105. doi:10.1088/1361-6501/ae65bf
BibTeX
@article{Fan2026,
author = {Fan, Zimeng and Du, Junrong and Jiao, Dongdong and Gui, Xuanang and Ning, Shijie and Song, Lei and Li, Xuzhi},
title = {Integrated fuzzy subdomain and test-time adaptation framework for cross-condition bearing RUL prediction},
journal = {Measurement Science and Technology},
year = {2026},
volume = {37},
number = {19},
pages = {195105},
doi = {10.1088/1361-6501/ae65bf}
}
EUVIMED