Predicting Endpoint Temperature of Molten Steel in VD Furnace Refining Process Using Metallurgical Mechanism and Bayesian Optimization XGBoost

Objective The vacuum degassing (VD) furnace is an essential piece of equipment for producing high-quality steel. Unlike other furnaces, the VD furnace lacks a heating function, leading to a significant drop in the temperature of molten steel during the refining process. If the refining endpoint temp...

Full description

Saved in:
Bibliographic Details
Main Authors: Ji XU, Zicheng XIN, Mo LAN, Wenhui LIN, Bo ZHANG, Qing LIU
Format: Article
Language:English
Published: Editorial Department of Journal of Sichuan University (Engineering Science Edition) 2024-11-01
Series:工程科学与技术
Subjects:
Online Access:http://jsuese.scu.edu.cn/thesisDetails#10.12454/j.jsuese.202400014
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850031880272871424
author Ji XU
Zicheng XIN
Mo LAN
Wenhui LIN
Bo ZHANG
Qing LIU
author_facet Ji XU
Zicheng XIN
Mo LAN
Wenhui LIN
Bo ZHANG
Qing LIU
author_sort Ji XU
collection DOAJ
description Objective The vacuum degassing (VD) furnace is an essential piece of equipment for producing high-quality steel. Unlike other furnaces, the VD furnace lacks a heating function, leading to a significant drop in the temperature of molten steel during the refining process. If the refining endpoint temperature of molten steel is excessively high, it results in energy waste and can even disrupt the continuous casting process. If the endpoint temperature is too low, the molten steel must be reheated in the ladle furnace (LF), severely impacting production efficiency. Therefore, the refining endpoint temperature significantly influences the quality of molten steel, production efficiency, and continuous casting. In recent years, adopting new technologies such as machine learning and big data has become crucial for transforming and upgrading steelmaking plants, facilitating the shift towards intelligent manufacturing. This study proposes a method that combines metallurgical mechanism analysis, data analysis, and machine learning to develop a temperature prediction model for VD furnace refining. Methods Initially, the key factors influencing the endpoint temperature of molten steel are identified through metallurgical mechanism analysis of the VD furnace. Then, production data from the steelmaking plant are processed following operational procedures to eliminate missing and abnormal data. The 3<italic>σ</italic> principle is applied to preprocess the actual production data, removing outliers to improve data quality and ensure the accuracy of subsequent model training and prediction. Pearson correlation analysis is then employed to discard factors with a minor impact on the molten steel temperature, helping to determine the input variables for the prediction model. The dataset is randomly shuffled, with 80% selected as the training set and the remaining 20% as the testing set. Following this, based on metallurgical mechanism (MM) analysis, the initial feature importance in the input variables of the extreme gradient boosting (XGBoost) model is partially amplified. Finally, the model’s prediction accuracy is further enhanced by optimizing the hyperparameters of XGBoost through Bayesian optimization (BO) algorithms, resulting in the development of MM–BO–XGBoost models. Results and Discussions This study employs grid search (GS) and random search (RS) for hyperparameter optimization of the model to compare the results to BO hyperparameter optimization. The prediction of molten steel temperature is performed using the metallurgical mechanism model, multiple linear regression model, and back-propagation neural network model as comparisons, demonstrating the superior performance of the MM–BO–XGBoost model. The results indicated that BO hyperparameter optimization is the most effective, providing the model with the best performance and higher prediction accuracy than the other two optimization algorithms. In addition, this model also exhibits good training efficiency, with the training time being slightly higher than RS and significantly shorter than that of GS. The optimal hyperparameter combination is determined to be max_depth = 7, learning_rate = 0.05, n_estimators = 345, with a model training time of 16.84 seconds. The fluctuations in the actual VD furnace refining endpoint molten steel temperature similarly impact the prediction errors of the MM–BO–XGBoost model and the other three existing models. Among the heats with endpoint molten steel temperatures below 1530 ℃, the mechanistic model predicts more heats with high molten steel temperature values, while the other three models predict fewer such heats. However, when the endpoint prediction temperature of molten steel exceeds 1530 ℃, the prediction accuracy of the mechanistic model diminishes, with many heats having prediction errors exceeding ±10 ℃, and a greater number exceeding ±20 ℃. The MM–BO–XGBoost model performs best, achieving the highest <italic>R</italic><sup>2</sup> of 0.909, the lowest RMSE of 6.741 ℃, and the lowest MAE of 5.114 ℃. The hit ratios of this model are 87.81% and 96.42% for ranges of [–10,10] and [–15,15], respectively. The MM–BO–XGBoost model demonstrates excellent training efficiency and prediction performance.Conclusion The analysis of the mechanism of the VD furnace refining process and Pearson correlation analysis are conducted, and the input variables for the prediction model of VD furnace endpoint liquid steel temperature are identified. Following the metallurgical mechanism, the characteristic importance of XGBoost input variables is partially enhanced, and the model’s hyperparameters are optimized using Bayesian optimization. Therefore, the MM–BO–XGBoost model is developed with strong applicability and high accuracy. This model is practical for precisely controlling molten steel temperature at the end of the VD furnace, ensuring continuous casting progression, and enhancing production efficiency. The input variables for the MM–BO–XGBoost model are identified as follows: initial temperature of molten steel, amount of calcium wire added, high vacuum holding time, amount of aluminum wire added, static stirring time, feeding time after vacuum break, vacuum degree, weight of molten steel, vacuum pumping time, argon gas flow during vacuum pumping, argon gas flow during high vacuum maintenance, and ladle arrival waiting time. The MM–BO–XGBoost model exhibits the best prediction performance of hyperparameters obtained through the Bayesian optimization algorithm. Compared to the traditional metallurgical mechanism, multiple linear regression, and BP neural network models, the MM–BO–XGBoost model in this study shows the best prediction effect.
format Article
id doaj-art-eb5f95a26ff54f2f9b2f83d60438ac2a
institution DOAJ
issn 2096-3246
language English
publishDate 2024-11-01
publisher Editorial Department of Journal of Sichuan University (Engineering Science Edition)
record_format Article
series 工程科学与技术
spelling doaj-art-eb5f95a26ff54f2f9b2f83d60438ac2a2025-08-20T02:58:51ZengEditorial Department of Journal of Sichuan University (Engineering Science Edition)工程科学与技术2096-32462024-11-0156637260040517Predicting Endpoint Temperature of Molten Steel in VD Furnace Refining Process Using Metallurgical Mechanism and Bayesian Optimization XGBoostJi XUZicheng XINMo LANWenhui LINBo ZHANGQing LIUObjective The vacuum degassing (VD) furnace is an essential piece of equipment for producing high-quality steel. Unlike other furnaces, the VD furnace lacks a heating function, leading to a significant drop in the temperature of molten steel during the refining process. If the refining endpoint temperature of molten steel is excessively high, it results in energy waste and can even disrupt the continuous casting process. If the endpoint temperature is too low, the molten steel must be reheated in the ladle furnace (LF), severely impacting production efficiency. Therefore, the refining endpoint temperature significantly influences the quality of molten steel, production efficiency, and continuous casting. In recent years, adopting new technologies such as machine learning and big data has become crucial for transforming and upgrading steelmaking plants, facilitating the shift towards intelligent manufacturing. This study proposes a method that combines metallurgical mechanism analysis, data analysis, and machine learning to develop a temperature prediction model for VD furnace refining. Methods Initially, the key factors influencing the endpoint temperature of molten steel are identified through metallurgical mechanism analysis of the VD furnace. Then, production data from the steelmaking plant are processed following operational procedures to eliminate missing and abnormal data. The 3<italic>σ</italic> principle is applied to preprocess the actual production data, removing outliers to improve data quality and ensure the accuracy of subsequent model training and prediction. Pearson correlation analysis is then employed to discard factors with a minor impact on the molten steel temperature, helping to determine the input variables for the prediction model. The dataset is randomly shuffled, with 80% selected as the training set and the remaining 20% as the testing set. Following this, based on metallurgical mechanism (MM) analysis, the initial feature importance in the input variables of the extreme gradient boosting (XGBoost) model is partially amplified. Finally, the model’s prediction accuracy is further enhanced by optimizing the hyperparameters of XGBoost through Bayesian optimization (BO) algorithms, resulting in the development of MM–BO–XGBoost models. Results and Discussions This study employs grid search (GS) and random search (RS) for hyperparameter optimization of the model to compare the results to BO hyperparameter optimization. The prediction of molten steel temperature is performed using the metallurgical mechanism model, multiple linear regression model, and back-propagation neural network model as comparisons, demonstrating the superior performance of the MM–BO–XGBoost model. The results indicated that BO hyperparameter optimization is the most effective, providing the model with the best performance and higher prediction accuracy than the other two optimization algorithms. In addition, this model also exhibits good training efficiency, with the training time being slightly higher than RS and significantly shorter than that of GS. The optimal hyperparameter combination is determined to be max_depth = 7, learning_rate = 0.05, n_estimators = 345, with a model training time of 16.84 seconds. The fluctuations in the actual VD furnace refining endpoint molten steel temperature similarly impact the prediction errors of the MM–BO–XGBoost model and the other three existing models. Among the heats with endpoint molten steel temperatures below 1530 ℃, the mechanistic model predicts more heats with high molten steel temperature values, while the other three models predict fewer such heats. However, when the endpoint prediction temperature of molten steel exceeds 1530 ℃, the prediction accuracy of the mechanistic model diminishes, with many heats having prediction errors exceeding ±10 ℃, and a greater number exceeding ±20 ℃. The MM–BO–XGBoost model performs best, achieving the highest <italic>R</italic><sup>2</sup> of 0.909, the lowest RMSE of 6.741 ℃, and the lowest MAE of 5.114 ℃. The hit ratios of this model are 87.81% and 96.42% for ranges of [–10,10] and [–15,15], respectively. The MM–BO–XGBoost model demonstrates excellent training efficiency and prediction performance.Conclusion The analysis of the mechanism of the VD furnace refining process and Pearson correlation analysis are conducted, and the input variables for the prediction model of VD furnace endpoint liquid steel temperature are identified. Following the metallurgical mechanism, the characteristic importance of XGBoost input variables is partially enhanced, and the model’s hyperparameters are optimized using Bayesian optimization. Therefore, the MM–BO–XGBoost model is developed with strong applicability and high accuracy. This model is practical for precisely controlling molten steel temperature at the end of the VD furnace, ensuring continuous casting progression, and enhancing production efficiency. The input variables for the MM–BO–XGBoost model are identified as follows: initial temperature of molten steel, amount of calcium wire added, high vacuum holding time, amount of aluminum wire added, static stirring time, feeding time after vacuum break, vacuum degree, weight of molten steel, vacuum pumping time, argon gas flow during vacuum pumping, argon gas flow during high vacuum maintenance, and ladle arrival waiting time. The MM–BO–XGBoost model exhibits the best prediction performance of hyperparameters obtained through the Bayesian optimization algorithm. Compared to the traditional metallurgical mechanism, multiple linear regression, and BP neural network models, the MM–BO–XGBoost model in this study shows the best prediction effect.http://jsuese.scu.edu.cn/thesisDetails#10.12454/j.jsuese.202400014VD furnace refiningsteel temperature predictionmechanism analysisMM–BO–XGBoost model
spellingShingle Ji XU
Zicheng XIN
Mo LAN
Wenhui LIN
Bo ZHANG
Qing LIU
Predicting Endpoint Temperature of Molten Steel in VD Furnace Refining Process Using Metallurgical Mechanism and Bayesian Optimization XGBoost
工程科学与技术
VD furnace refining
steel temperature prediction
mechanism analysis
MM–BO–XGBoost model
title Predicting Endpoint Temperature of Molten Steel in VD Furnace Refining Process Using Metallurgical Mechanism and Bayesian Optimization XGBoost
title_full Predicting Endpoint Temperature of Molten Steel in VD Furnace Refining Process Using Metallurgical Mechanism and Bayesian Optimization XGBoost
title_fullStr Predicting Endpoint Temperature of Molten Steel in VD Furnace Refining Process Using Metallurgical Mechanism and Bayesian Optimization XGBoost
title_full_unstemmed Predicting Endpoint Temperature of Molten Steel in VD Furnace Refining Process Using Metallurgical Mechanism and Bayesian Optimization XGBoost
title_short Predicting Endpoint Temperature of Molten Steel in VD Furnace Refining Process Using Metallurgical Mechanism and Bayesian Optimization XGBoost
title_sort predicting endpoint temperature of molten steel in vd furnace refining process using metallurgical mechanism and bayesian optimization xgboost
topic VD furnace refining
steel temperature prediction
mechanism analysis
MM–BO–XGBoost model
url http://jsuese.scu.edu.cn/thesisDetails#10.12454/j.jsuese.202400014
work_keys_str_mv AT jixu predictingendpointtemperatureofmoltensteelinvdfurnacerefiningprocessusingmetallurgicalmechanismandbayesianoptimizationxgboost
AT zichengxin predictingendpointtemperatureofmoltensteelinvdfurnacerefiningprocessusingmetallurgicalmechanismandbayesianoptimizationxgboost
AT molan predictingendpointtemperatureofmoltensteelinvdfurnacerefiningprocessusingmetallurgicalmechanismandbayesianoptimizationxgboost
AT wenhuilin predictingendpointtemperatureofmoltensteelinvdfurnacerefiningprocessusingmetallurgicalmechanismandbayesianoptimizationxgboost
AT bozhang predictingendpointtemperatureofmoltensteelinvdfurnacerefiningprocessusingmetallurgicalmechanismandbayesianoptimizationxgboost
AT qingliu predictingendpointtemperatureofmoltensteelinvdfurnacerefiningprocessusingmetallurgicalmechanismandbayesianoptimizationxgboost