MCA-Mamba-2: A Multi-Channel Attention State-Space Network for Ultrasonic Guided Wave Damage Detection in Small-Diameter Pipes
Qiuyu Yu, Ruohua Zhou, Zhengxian Liang, Yajun Liu · 2026
Read MCA-Mamba-2: A Multi-Channel Attention State-Space Network for Ultrasonic Guided Wave Damage Detection in Small-Diameter Pipes by Qiuyu Yu, Ruohua Zhou, Zhengxian Liang, Yajun 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/pmc-13611702.txt.
Description
Pipelines are checked for hidden damage by sending ultrasound waves along the pipe wall, but small pipes give a narrow band of usable frequencies and noisy readings. Qiuyu Yu and co-authors built a neural network for the job, combining a convolutional front end with a bidirectional state-space backbone and two kinds of attention. The network listens to four waveforms at once from a ring of receivers around the pipe.
On a 20-millimeter steel pipe with ten damage states, from surface abrasions to through-wall holes. The model hit 98 percent accuracy on clean signals and stayed above 81 percent under heavy added noise, beating its backbone alone in every noise test. The authors add caveats: accuracy collapsed when the noise type differed from training, though training on mixed noise recovered much of it, and all data came from one pipe.
Catalog record
| CREATOR | Qiuyu Yu, Ruohua Zhou, Zhengxian Liang, Yajun Liu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 61 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |