BEGIN:VCALENDAR
VERSION:2.0
METHOD:PUBLISH
BEGIN:VEVENT
ORGANIZER;CN=ESTAD 2023:mailto:info@metec-estad.com
LOCATION:Room 26
SUMMARY:Development of numerical model based deep learning for the roll force prediction in the sendzimir mill
DESCRIPTION:The ZRM (Sendzmir Mill) is the cold rolling mill which consists of 20 rolls to produce the stainless steels and the electrical steels. The work roll of the ZRM is so small to reduce the thickness of the strip efficiently. At the beginning of the rolling, the roll force set-up was performed for the desired thickness of the strip. If the set value of the roll force is not correct for the outlet thickness, the thickness deviation is high, furthermore the strip is broken during rolling. For the ZRM process, sound prediction of the roll force is vital for achieving the desired thickness because the stability of the rolling substantially affected by it.
In this paper, mathematical model is presented for the prediction of the roll force at the beginning of the rolling. The model consists of a numerical model for the prediction of the roll force, a sub-model for the prediction of the mechanical property of the strip by the deep learning, which is the deep neural networks. It is called numerical model based deep learning. From the combination of these models, the roll force at the beginning of the rolling can be predicted to produce the desired thickness of the strip.
The numerical model based deep learning has several advantages compared with typical deep learning. The prediction accuracy has been improved from the physical tendency of the numerical model. In addition, from reduced number of inputs of the deep learning model, the learning time was decreased. The prediction accuracy of the proposed model is examined through comparison with actual data.
CLASS:PUBLIC
DTSTART:20230614T090000
DTEND:20230614T092000
END:VEVENT
END:VCALENDAR