EUVIMEDEUVIMED entdecken
Crossref

Deep learning based automatic analysis of multimodal TFM data in elbow-tube weld junctions in nuclear

Roberto Miorelli, David Roué, Sébastien Robert

e-Journal of Nondestructive Testing · 2026-07 · 31 (7)

DOI: 10.58286/33343 ↗

Kurz gefasst

Fachartikel in der Zeitschrift e-Journal of Nondestructive Testing, 2026-07. Verfasst von Roberto Miorelli und 2 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper presents a study aimed at detecting defect echoes in TFM images while using Machine Learning (ML) algorithms as integrated support decision tool for Non-Destructive Testing (NDT) experts. The application case identified is the inspection of weld pipe-elbow (P-E) junction surroundings in RIS (Reactor Safety Injection) and RRA (Reactor Shutdown Cooling) lines for the detection and characterization of Stress Corrosion Cracking (SCC). The data are issue from multiple experimental campaigns relying based on Multimodal Total Focusing Method (M-TFM) acquisitions conduct or realistic inspection set-ups. This work relies on the use of Deep Learning (DL) pipeline specifically tailored for the discrimination between artifacts and defect echoes in M-TFM imaging despite uncertainties about the component's geometry

Zitieren

APA

Miorelli, R., Roué, D., & Robert, S. (2026). Deep learning based automatic analysis of multimodal TFM data in elbow-tube weld junctions in nuclear. e-Journal of Nondestructive Testing, 31(7). https://doi.org/10.58286/33343

Vancouver

Miorelli R, Roué D, Robert S. Deep learning based automatic analysis of multimodal TFM data in elbow-tube weld junctions in nuclear. e-Journal of Nondestructive Testing. 2026;31(7). doi:10.58286/33343

BibTeX
@article{Miorelli2026,
  author = {Miorelli, Roberto and Roué, David and Robert, Sébastien},
  title = {Deep learning based automatic analysis of multimodal TFM data in elbow-tube weld junctions in nuclear},
  journal = {e-Journal of Nondestructive Testing},
  year = {2026},
  volume = {31},
  number = {7},
  doi = {10.58286/33343}
}
Kostenlos starten →