Deployment of AI-Assisted Ultrasonic Inspections in the Nuclear Power Industry
e-Journal of Nondestructive Testing · 2026-07 · 31 (7)
DOI: 10.58286/33295 ↗Kurz gefasst
Fachartikel in der Zeitschrift e-Journal of Nondestructive Testing, 2026-07. Verfasst von Thiago Osorio.
Abstract (Originalsprache)
Ultrasonic testing is the main volumetric inspection method employed in the nuclear power industry and forms an essential part of the non-destructive evaluation program of all plants. Many of the ultrasonic inspections performed in the field yield large volumes of data that need to be carefully reviewed by multiple qualified inspectors during an outage. Even though most of the data is benign, a detailed review of all data is needed, requiring inspectors to maintain high levels of focus for extended periods of time in an unfavorable environment. To provide an aid to the inspector in these analyses, EPRI have developed machine learning tools that perform automated analysis of the data to flag suspicious regions for further review and screen out benign regions, thus greatly reducing the volume of data requiring human analysis. Such tools are not something for the future; they are used in the field today. Indeed, one such tool for reactor vessel upper head penetrations has now been qualified and deployed in nuclear power plants. This presentation will discuss field implementations to date as well as additional applications under development and assessment.
Zitieren
APA
Osorio, T. (2026). Deployment of AI-Assisted Ultrasonic Inspections in the Nuclear Power Industry. e-Journal of Nondestructive Testing, 31(7). https://doi.org/10.58286/33295
Vancouver
Osorio T. Deployment of AI-Assisted Ultrasonic Inspections in the Nuclear Power Industry. e-Journal of Nondestructive Testing. 2026;31(7). doi:10.58286/33295
BibTeX
@article{Osorio2026,
author = {Osorio, Thiago},
title = {Deployment of AI-Assisted Ultrasonic Inspections in the Nuclear Power Industry},
journal = {e-Journal of Nondestructive Testing},
year = {2026},
volume = {31},
number = {7},
doi = {10.58286/33295}
}
EUVIMED