NO SHOW: Implementing Dimensionality Reduction for Fundamental Assisted Analytics for End of Blow Chemistry Prediction in an LD Converter (FAA4LD)

End-of-blow chemistry predictions are an important tool for LD process control and achieving high strike rates. Existing models are primarily based on thermodynamic/kinetic principles or by deploying advanced sensors like offgas analyzer/sublance systems. This study aims to combine the strengths of theoretical models and machine-learning techniques to improve the predictive capacity of the end-of-blow chemistry model (FAA4LD). The results show that FAA4LD can increase prediction accuracy for thermodynamic-based models by reducing the number of assumptions during the formulation; and improving interpretability and generalization for data-driven models, allowing them to be implemented universally.

Dr Soumitra Kumar Dinda | University of Toronto
 

Mr Ruibin Wang | University of Toronto
 

Dr Itishree Mohanty | Tata Steel Ltd.
 

Mr Prakash Gupta | Tata Steel Ltd.
 

Dr Tapas Kumar Roy | Tata Steel Ltd.