Rolling bearing fault diagnosis method based on dual attention multi-scale convolutional network
Yamei Zhang, Xiaohu Liu · 2026
Read Rolling bearing fault diagnosis method based on dual attention multi-scale convolutional network by Yamei Zhang, Xiaohu Liu (2026) 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/plos-pone-0358253.txt.
Description
Yamei Zhang and Xiaohu Liu build a neural network to hear failing bearings before they seize, in PLOS ONE in 2026. Rolling bearings support nearly every rotating machine, and their faults hide under noise and shifting speeds. Their network reads vibration signals at several scales at once. Then it applies two kinds of attention. One reweights which signal channels matter. The other spotlights the brief fault impulses in the stream. Only then does it classify the fault type.
On the standard public bearing dataset, the system reached accuracy above ninety-nine percent in the hardest cross condition tests, where the machine's speed and load change between training and testing. Under heavy added noise it still held above ninety-two percent. It beat three recent rival methods on accuracy and related scores while running cheaper and faster. For factory maintenance teams, that combination promises earlier warnings from the sensors they already own.
Catalog record
| CREATOR | Yamei Zhang, Xiaohu Liu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45302 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |