Model-based sticker detection in continuous casting
Résumé
Many innovations are currently initiated by scientific community in the field of dynamic system diagnosis. Arcelormittal Maizières R&D adapted and evaluated some advanced diagnosis tools for the detection of abnormal events and operating mode changing in steel processes. The developed approach is illustrated and applied to the sticker detection in continuous casting process, where the prevention of breakouts in slabs is a crucial challenge. Results show that it is possible to reduce relatively, the time delay for sticker detection and the false alarms rate without reducing the sticker detection rate. The evaluation of the proposed system on industrial data is encouraging and promising. This adaptable software solution could contribute significantly to improve the supervision and the productivity of the process with low or no investment.
Domaines
Automatique / Robotique
Origine : Fichiers produits par l'(les) auteur(s)