Visual Argon Stirring Measurement in Ladle Furnace Using Deep Learning

Having a value indicating the level of argon stirring of steel is important for process and quality control, but argon pressure and flow measurements are not strong and stable enough to obtain a stirring value. An artificial intelligence system was developed that uses images from thermal cameras. The system was trained through deep learning to recognize the degree of movement of the steel mass and thus classify the level of stirring for each pore of the ladle furnace. This signal becomes a useful trending displayed in L2 human-machine interface. The system began operation in February 2023 at Ternium Argentina.

Authors:

Omar Martin | Ternium Argentina S.A.
 

Ariel González | Janus Automation S.A.
 

Federico Avila | Janus Automation S.A.
 

Session Chairs:

Judy Li | Cleveland-Cliffs Research & Innovation Center
 

Morgan Ashbaugh | Opta (USA) Inc.