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

Abstract

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 optimization (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 prediction 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 exPlanations (SHAP) analysis. Consequently, the explainability of the optimal GWO-LGBM model was enhanced, providing reliable support for prediction results.

Keywords

ladle furnace refining / molten steel temperature / eXtreme gradient boosting / light gradient boosting machine / grey wolf optimization / SHapley Additive exPlanation

References

[1] Yin RY. A discussion on “smart” steel plant—View from physical system side. Iron Steel. 2017, 52(6): 1

[2] Xin ZC, Zhang JS, Jin Y, Zheng J, Liu Q. Predicting the alloying element yield in a ladle furnace using principal component analysis and deep neural network. Int. J. Miner. Metall. Mater.. 2023, 30(2): 335

[3] Liu Q, Shao X, Yang JP, Zhang JS. Multiscale modeling and collaborative manufacturing for steelmaking plants. Chin. J. Eng.. 2021, 43(12): 1698

[4] Xin ZC, Zhang JS, Peng KX, Zhang JG, Zhang CH, Liu Q. Modeling of LF refining process: A review. J. Iron Steel Res. Int.. 2024, 31(2): 289

[5] Samways NL, Dancy TE. Factors affecting temperature drop between tapping and teeming. JOM. 1960, 12(4): 331

[6] Zimmer A, Lima NC, Trommer RM, Bragança SR, Bergmann CP. Heat transfer in steelmaking ladle. J. Iron Steel Res. Int.. 2008, 15(3): 11

[7] Wu YJ, Jiang ZH, Jiang MF, Gong W, Zhan DP. Temperature prediction model of molten steel in LF. J. Iron Steel Res.. 2002, 14: 9

[8] Srinivas PS, Kothari AK, Agrawal A. Parameter estimation by inverse solution methodology using genetic algorithms for real time temperature prediction model of ladle furnace. ISIJ Int.. 2016, 56(6): 977

[9] Wang AN, Tian HX, Jiang ZH, Zhan DP, Yin XD, Ma ZG. Temperature prediction of molten steel for LF based on information fusion technique. J. Iron Steel Res.. 2005, 17(6): 71

[10] Xin ZC, Zhang JS, Zhang JG, Zheng J, Jin Y, Liu Q. Predicting temperature of molten steel in LF-refining process using IF–ZCA–DNN model. Metall. Mater. Trans. B. 2023, 54(3): 1181

[11] Lv W, Mao ZZ, Yuan P, Jia MX. Multi-kernel learnt partial linear regularization network and its application to predict the liquid steel temperature in ladle furnace. Knowl. Based Syst.. 2012, 36: 280

[12] Yuan F, Xu AJ, Gu MQ. Development of an improved CBR model for predicting steel temperature in ladle furnace refining. Int. J. Miner. Metall. Mater.. 2021, 28(8): 1321

[13] Tian HX, Liu YD, Li K, Yang RR, Meng B. A new AdaBoost.IR soft sensor method for robust operation optimization of ladle furnace refining. ISIJ Int.. 2017, 57(5): 841

[14] Xin ZC, Zhang JS, Zheng J, Jin Y, Liu Q. A hybrid modeling method based on expert control and deep neural network for temperature prediction of molten steel in LF. ISIJ Int.. 2022, 62(3): 532

[15] Çamdali Ü, Tunc M. Energy and exergy analysis of a ladle furnace. Can. Metall. Q.. 2003, 42: 439

[16] Tao ZY, Jiang MF, Liu CJ. Prediction of molten steel end point temperature in LF based on modified artificial neural network. Spec. Steel. 2006, 27(6): 21

[17] Fu GQ, Liu Q, Wang Z, et al. . Grey box model for predicting the LF end-point temperature of molten steel. J. Univ. Sci. Technol. Beijing. 2013, 35(7): 948

[18] L.L. Zou, J.S. Zhang, Y.S. Han, F.Z. Zeng, Q.H. Li, and Q. Liu, Internal crack prediction of continuous casting billet based on principal component analysis and deep neural network, Metals, 11(2021), No. 12, art. No. 1976.

[19] Xu G, Li M, Lv ZM, Xu JW. Online intelligent product quality monitoring method based on machine learning. Chin. J. Eng.. 2022, 44(4): 730

[20] Shao X, Liu Q, Xin ZC, Zhang JS, Zhou T, Li SS. Hybrid model for BOF oxygen blowing time prediction based on oxygen balance mechanism and deep neural network. Int. J. Miner. Metall. Mater.. 2024, 31(1): 106

[21] Niu Y, Hong ZQ, Wang YQ, Zhu YC. Machine learning-based beta transus temperature prediction for titanium alloys. J. Mater. Res. Technol.. 2023, 23: 515

[22] Shafighfard T, Bagherzadeh F, Rizi RA, Yoo DY. Data-driven compressive strength prediction of steel fiber reinforced concrete (SFRC) subjected to elevated temperatures using stacked machine learning algorithms. J. Mater. Res. Technol.. 2022, 21: 3777

[23] P. Mahajan, S. Uddin, F. Hajati, and M.A. Moni, Ensemble learning for disease prediction: A review, Healthcare, 11(2023), No. 12, art. No. 1808.

[24] Li HY, Liu XJ, Li X, Bu XP, Li HW, Lyu Q. Causality analysis and prediction of blast furnace state based on convergence cross mapping. Ironmaking Steelmaking. 2022, 49(9): 875

[25] Y.K. Kong and K. Kurumisawa, Application of machine learning in predicting workability for alkali-activated materials, Case Stud. Constr. Mater., 18(2023), art. No. e02173.

[26] Pelikan M, Goldberg DE, Cantú-Paz E. BOA: The Bayesian optimization algorithm. Proceedings of the Genetic and Evolutionary Computation Conference. 1999525Orlando

[27] H. Huang, J.N. Qiu, and K. Riedl, On the global convergence of particle swarm optimization methods, Appl. Math. Optim., 88(2023), No. 2, art. No. 30.

[28] Holland JH. Genetic algorithms. Sci. Am.. 1992, 267(1): 66

[29] Mirjalili S, Mirjalili SM, Lewis A. Grey wolf optimizer. Adv. Eng. Software. 2014, 69: 46

[30] Cui RF, Ma H, Guo CP, Li HY, Liu Z. Prediction of missile aerodynamic data based on gradient boosting under Bayesian hyperparametric optimization. Aeronaut. Sci. Technol.. 2023, 34(7): 22

[31] B. Gong, Study of PLSR-BP model for stability assessment of loess slope based on particle swarm optimization, Sci. Rep., 11(2021), No. 1, art. No. 17888.

[32] He YB, Guo H, Zhang BQ, Zhu Q, Tang HM, Li YH. Prediction method of hot metal silicon content based on improved BP neural network. J. Iron. Steel. Res.. 2024, 36(3): 309

[33] Zhou J, Zhang YL, Li CQ, et al. . Enhancing the performance of tunnel water inflow prediction using Random Forest optimized by Grey Wolf Optimizer. Earth Sci. Inform.. 2023, 16(3): 2405

[34] Ribeiro MT, Singh S, Guestrin C. “Why should I trust you?”: Explaining the predictions of any classifier. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 20161135 San Diego

[35] Lundberg SM, Lee SI. A unified approach to interpreting model predictions. Proceedings of the 31st Conference and Workshop on Neural Information Processing Systems. 20174765California

[36] Huang YL, Qin XL, Chen YW, Zhang LG, Yi B. Interpretability analysis of sepsis prediction model using LIME. J. Comput. Appl.. 2021, 41(S1): 332

[37] Bar N, Korem T, Weissbrod O, et al. . A reference map of potential determinants for the human serum metabolome. Nature. 2020, 588(7836): 135

[38] W. Liu, Z.X. Chen, Y. Hu, and J. Zhang, Forecasting pipeline safety and remaining life with machine learning methods and SHAP interaction values, Int. J. Press. Vessels Pip., 205(2023), art. No. 105000.

[39] A. Homafar, H. Nasiri, and S.C. Chelgani, Modeling coking coal indexes by SHAP-XGBoost: Explainable artificial intelligence method, Fuel Commun., 13(2022), art. No. 100078.

[40] Ma SL. Theoretical analysis and calculation of LF fume parameters. Environ. Eng.. 2007, 25(5): 44

[41] Tian HX, Mao ZZ, Wang AN. Hybrid modeling for soft sensing of molten steel temperature in LF. J. Iron Steel Res. Int.. 2009, 16(4): 1

[42] Li J, Fu J, Wang P, Bi QF, Huang CG. Composition adjustment and temperature prediction of molten steel in LF. J. Iron Steel Res.. 1999, 11(2): 6

[43] S.M. Malakouti, Babysitting hyperparameter optimization and 10-fold-cross-validation to enhance the performance of ML methods in predicting wind speed and energy generation, Intell. Syst. Appl., 19(2023), art. No. 200248.

[44] Taylor KE. Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res.. 2001, 106(D7): 7183

[45] S. Mangalathu, S.H. Hwang, and J.S. Jeon, Failure mode and effects analysis of RC members based on machine-learning-based SHapley Additive exPlanations (SHAP) approach, Eng. Struct., 219(2020), art. No. 110927.

[46] Xi B, Li EM, Fissha Y, Zhou J, Segarra P. LGBM-based modeling scenarios to compressive strength of recycled aggregate concrete with SHAP analysis. Mech. Adv. Mater. Struct.. 2023, 31(23): 5999