Application of Machine Learning for the Reduction of Cobbles in Hot Strip Mills

The hot rolling mill process is a pivotal stage in steel production, where hot steel slabs are passed through rollers to achieve desired thickness, shape and material property. However, a persistent challenge in this process is the occurrence of cobbles. With the help of machine learning, this research develops a Cobble Prediction and Analysis model which shows the mill operators a probability score that a strip will cobble and shows the process engineers the root causes for the cobbles. This model helps to reduce the mill downtime, avoid equipment damage and increase the rollability of the cobble-prone products.

Authors:

Srikanta Kolay | Big River Steel – A U. S. Steel Co.
 

Aaron Meyer | Big River Steel – A U. S. Steel Co.
 

Thouraya Ben Kilani | Big River Steel – A U. S. Steel Co.
 

Session Chairs:

Ashish Gupta | Big River Steel
 

Andrei Chastukhin | Mississippi State University Center for Advanced Vehicular Systems