CIEMAT Publications
ISSN: 2695-8864 | NIPO: 152-24-006-6
PUBLISHED IN 2024
Informes técnicos CIEMAT
Herramientas para la detección de MARFE en JET con cámaras operacionales
González Ganzábal, A.; Rattá, G.A.; Gadariya, D.; Dormido Canto, S.
Free
Abstract
A set of tools have been developed to detect MARFEs using the videos from JET's operational cameras. Firstly, three of these tools work directly with information extracted from the videos, obtaining precisions of up to 91.4%. With that same information, machine learning tools have been created, including a final model with additional JET signals and diagnostics, as well as temperature and electron density profiles, with a detection accuracy of 96.9%. After obtaining notable results, the machine learning methods were adapted to distinguish between MARFEs and soft-landed MARFEs, reaching a score of 98% for the full model. In addition, various tools are presented for downloading and synchronizing JET videos and signals, as well as PETRA alarms.