Coke Strength Prediction With Neural Networks

This study explores the effectiveness of neural networks in predicting coke strength after reaction (CSR) compared to classical regression models. CSR is a crucial parameter in ironmaking, influencing blast furnace efficiency. Traditional regression struggles with complex CSR prediction due to linear limitations, while neural networks excel in grasping intricate patterns in extensive datasets including raw material quality and composition, process conditions, and other factors. This research compares neural networks to classical regression models, demonstrating their accuracy in predicting CSR, which could revolutionize the ironmaking industry. The findings suggest a potential paradigm shift toward adopting neural networks for improved CSR forecasting.

Authors:

Dr. Andreas Schwabauer | thyssenkrupp Steel Europe AG
 

Viktor Stiskala | thyssenkrupp Steel Europe AG
 

Session Chairs:

Steve McKnight | U. S. Steel - Clairton Works
 

James Hosfield | U. S. Steel - Clairton Works
 

Dick Randolph | Fosbel Inc.