Explainable machine learning model for predicting molten steel temperature in the LF refining process

Accurate prediction of molten steel temperature in the ladle furnace(LF)refining process has an important influence on the quality of molten steel and the control of steelmaking cost.Extensive research on establishing models to predict molten steel temperature has been conducted.However,most researchers focus solely on improving the accuracy of the model,neglecting its explainability.The present study aims to develop a high-precision and explainable model with improved reliability and transparency.The eXtreme gradient boosting(XGBoost)and light gradient boosting machine(LGBM)were utilized,along with bayesian optimization and grey wolf optimiz-ation(GWO),to establish the prediction model.Different performance evaluation metrics and graphical representations were applied to compare the optimal XGBoost and LGBM models obtained through varying hyperparameter optimization methods with the other models.The findings indicated that the GWO-LGBM model outperformed other methods in predicting molten steel temperature,with a high pre-diction accuracy of 89.35%within the error range of±5°C.The model’s learning/decision process was revealed,and the influence degree of different variables on the molten steel temperature was clarified using the tree structure visualization and SHapley Additive exPlana-tions(SHAP)analysis.Consequently,the explainability of the optimal GWO-LGBM model was enhanced,providing reliable support for prediction results.

Influence of refining process and utilization of different slags on inclusions, titanium yield and total oxygen content of Ti-stabilized 321 stainless steel

Ti-stabilized 321 stainless steel was prepared using an electric arc furnace, argon oxygen decarburization (AOD) furnace, ladle furnace (LF), and continuous casting processes. In addition, the effect of refining process and utilization of different slags on the evolution of inclusions, titanium yield, and oxygen content was systematically investigated by experimental and thermodynamic analysis. The results reveal that the total oxygen content (TO) and inclusion density decreased during the refining process. The spherical CaO–SiO2–Al2O3–MgO inclusions existed in the 321 stainless steel after the AOD process. Moreover, prior to the Ti addition, the spherical CaO–Al2O3–MgO–SiO2 inclusions were observed during LF refining pro-cess. However, Ti addition resulted in multilayer CaO–Al2O3–MgO–TiOx inclusions. Two different samples were prepared by conventional CaO–Al2O3-based slag (Heat-1) and -TiO2-rich CaO–Al2O3-based slag (Heat-2). The statistical analysis revealed that the density of inclusions and the -TiOx content in CaO–Al2O3–MgO–TiOx inclusions found in Heat-2 sample are much lower than those in the Heat-1 sample. Furthermore, the TO content and Ti yield during the LF refining process were controlled by using -TiO2-rich calcium aluminate synthetic slag. These results were consistent with the ion–molecule coexist-ence theory and FactSage?7.2 software calculations. When -TiO2-rich CaO–Al2O3-based slag was used, the -TiO2 activity of the slag increased, and the equilibrium oxygen content significantly decreased from the AOD to LF processes. Therefore, the higher -TiO2 activity of slag and lower equilibrium oxygen content suppressed the undesirable reactions between Ti and O.