Integrating Frequency Guidance into Multi-source Domain Generalization for Bearing Fault Diagnosis Journal: Advanced Engineering Informatics
Xiaotong Tu, Chenyu Ma, Qingyao Wu, Yinhao Liu, Hongyang Zhang · 2025
Read Integrating Frequency Guidance into Multi-source Domain Generalization for Bearing Fault Diagnosis Journal: Advanced Engineering Informatics by Xiaotong Tu, Chenyu Ma, Qingyao Wu, Yinhao Liu, Hongyang Zhang (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-00545.txt.
Description
Xiaotong Tu and co-authors teach a neural network to spot bearing faults under working conditions it has never seen. Fault diagnosis models often fail when conditions shift, because features tied to one setting carry spurious links to the fault labels. Their network, FARNet, works in the frequency domain, where a signal splits into phase and amplitude parts that hold different kinds of information.
One sub-network handles the amplitude spectrum and another the phase, jointly shrinking the gap between source and target settings. A frequency-spatial module mixes global information with local detail. Training uses augmented data from several sources, and a new manifold triplet loss sharpens the decision boundary. On two standard bearing benchmarks, the model beat current cross-domain rivals. The authors present frequency guidance as a practical route to diagnosis systems that keep working when machines move into new conditions.
Catalog record
| CREATOR | Xiaotong Tu, Chenyu Ma, Qingyao Wu, Yinhao Liu, Hongyang Zhang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-00545