A cross-domain geometric-temporal correlation network for bearing fault diagnosis under varying operating conditions.

Authors

Jing Zhu, Shilei Liang, Ting Liao, Minghui Chen, Siyuan Cao

Published in

ISA transactions. Nov 02, 2026. Epub Nov 02, 2026.

Abstract

Under varying speed and load conditions, vibration signals often exhibit nonstationary responses and spectral energy redistribution, which can reorganize latent fault representations and degrade diagnostic performance under unlabeled target operating conditions. Existing cross-domain diagnosis methods mainly rely on statistical distribution alignment, but they do not explicitly account for condition-induced representation reorganization, which may limit the stability of learned fault representations across operating conditions. To address this issue, this paper proposes a Cross-domain Geometric-Temporal Correlation Network (CrossGTCN) for cross-condition bearing fault diagnosis from a representation-stability perspective. CrossGTCN coordinates relational modeling, cross-view interaction, and geometry-aware discrimination to improve the stability of condition-sensitive representations. Specifically, a topology-conditioned temporal modulation mechanism converts graph-derived neighborhood information into feature-wise modulation parameters to recalibrate temporal representations, while a frequency-domain energy-relation branch captures complementary spectral relationships associated with multi-band energy redistribution. The resulting temporal-structural and frequency-domain representations are further coordinated through a mutual-evidence cross-view fusion mechanism, and the fused representation is regularized in an angular space to preserve class-discriminative geometry across operating conditions. Experiments on 20 transfer tasks across the JNU, HUST, and SEU datasets show that CrossGTCN achieves an average diagnostic accuracy above 98%. Ablation and visualization results further demonstrate the effectiveness of the proposed framework in improving representation stability and cross-condition discriminability.

PMID:
42995130
Bibliographic data and abstract were imported from PubMed on 07 Nov 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.