Intelligent Prediction for Endpoint Mass Fraction of Carbon in Molten Steel of RH

doi:10.12068/j.issn.1005-3026.2025.20240051

Abstract

Abstract:

Accurate prediction of the endpoint mass fraction of carbon in the molten steel of RH （Ruhrstahl Heraeus） can effectively improve the quality of continuously cast products. In order to realize this goal， data mining was firstly applied to preprocess the RH industrial data. Then， grey correlation analysis， Spearman correlation coefficient， and random forest out-of-bag error scoring were used to select the features that had a strong correlation with the endpoint mass fraction of carbon in the molten steel. Next， the principal component analysis method was applied to reduce the dimensions. Finally， the XGBoost model， the XGBoost model optimized by the particle swarm optimization algorithm， and the XGBoost model optimized by the whale optimization algorithm were applied to predict the endpoint mass fraction of carbon in the molten steel. The results show that grey correlation analysis is better than Spearman rank correlation coefficient and random forest in analyzing the selected features. After the optimization of the particle swarm optimization algorithm and whale optimization algorithm， the XGBoost model has a greater prediction hit rate. The XGBoost model optimized by the whale optimization algorithm is better than that by the particle swarm optimization algorithm. In the case of the XGBoost model optimized by the whale optimization algorithm， the hit rate reaches 91.26% and 98.97% if the error range is within ±5×10^-6， and ±7×10^-6.

Key words: endpoint mass fraction of carbon in molten steel, XGBoost algorithm, particle swarm optimization algorithm, whale optimization algorithm, feature selection

CLC Number:

TF 769.4

Deng-hui LI, Yan ZHAO, Hong LEI, Jia FAN. Intelligent Prediction for Endpoint Mass Fraction of Carbon in Molten Steel of RH[J]. Journal of Northeastern University(Natural Science), 2025, 46(10): 51-58.

Figures/Tables 8

References 24

[1]	饶江平，杨治争，李光强，等. IF钢RH精炼理论研究与工艺优化［J］. 炼钢， 2022， 38（5）： 59-66.
	Rao Jiang-ping， Yang Zhi-zheng， Li Guang-qiang， et al. Theoretical study and process optimization of RH refining for IF steel［J］. Steelmaking， 2022， 38（5）： 59-66.
[2]	Feng K， Xu A J， Wu P F， et al. Case-based reasoning model based on attribute weights optimized by genetic algorithm for predicting end temperature of molten steel in RH［J］. Journal of Iron and Steel Research International， 2019， 26（6）： 585-592.
[3]	Lu W， Mao Z Z， Yuan P. Ladle furnace liquid steel temperature prediction model based on optimally pruned bagging［J］. Journal of Iron and Steel Research International， 2012， 19（12）： 21-28.
[4]	Chen G J， Yang J， Li Y G， et al. Mathematical simulation of decarburization with CO₂ injection during RH refining of ultra-low-carbon steel［J］. Metals and Materials International， 2025， 31： 167-181.
[5]	Li Y W， Liu B G， Peng J H， et al. Prediction model of microwave calcining of ammonium diuranate using incremental improved back-propagation neural network［J］. Acta Metallugica Sinica（English Letters）， 2011， 24（1）： 34-42.
[6]	Wang S F， Tang Y， Li X B， et al. Analyses and predictions of rock cuttabilities under different confining stresses and rock properties based on rock indentation tests by conical pick［J］. Transactions of Nonferrous Metals Society of China， 2021， 31（6）： 1766-1783.
[7]	Lee E H， Kim K， Kho S Y， et al. Estimating express train preference of urban railway passengers based on extreme gradient boosting （XGBoost） using smart card data［J］. Transportation Research Record， 2021， 2675： 64-76.
[8]	刘志明，战东平，葛启桢，等. 基于BP神经网络的电炉终点碳质量分数预报模型［J］. 工业加热， 2018， 47（4）： 28-31.
	Liu Zhi-ming， Zhan Dong-ping， Ge Qi-zhen， et al. Prediction model of mass fraction of endpoint carbon of electric furnace based on BP neural network［J］. Industrial Heating， 2018， 47（4）： 28-31.
[9]	魏付豪，刘建华，张游游，等. RH精炼终点预报模型［J］. 炼钢， 2016， 32（6）： 38-44.
	Wei Fu-hao， Liu Jian-hua， Zhang You-you， et al. The endpoint prediction model for RH refining［J］. Steelmaking， 2016， 32（6）： 38-44.
[10]	杨业鹏，岳峰，马明胜. RH精炼炉脱碳模型研究［J］. 炼钢， 2020， 36（2）： 10-16.
	Yang Ye-peng， Yue Feng， Ma Ming-sheng. Study on decarburization model for RH refining furnace［J］. Steelmaking， 2020， 36（2）： 10-16.
[11]	Heo J， Kim T W， Jung S J， et al. Real-time prediction model of carbon content in RH process［J］. Applied Sciences， 2022， 12（21）： 10753-10764.
[12]	陈超，农伟民，王楠. 基于机器学习模型的Consteel电弧炉终点碳含量及温度预测［J］. 冶金自动化， 2023， 47（6）： 37-44.
	Chen Chao， Nong Wei-min， Wang Nan. Prediction on end-point carbon content and temperature of Consteel electric arc furnace based on machine learning model［J］. Metallurgical Industry Automation， 2023， 47（6）： 37-44.
[13]	Sun Y， Brown M B， Prapopoulou M， et al. The application of stochastic machine learning methods in the prediction of skin penetration［J］. Applied Soft Computing， 2011， 11（2）： 2367-2375.
[14]	Achour A， Kammoun M A， Hajej Z. Towards optimizing multi-level selective maintenance via machine learning predictive models［J］. Applied Sciences， 2024， 14（1）： 313-318.
[15]	Qu Z， Genton M G. Sparse functional boxplots for multivariate curves［J］. Journal of Computational and Graphical Statistics， 2022， 31（4）： 976-989.
[16]	Er O， Külekci M K， Esme U， et al. Multi response optimization of friction stir spot welding process using Taguchi based grey relational analysis［J］. Cukurova University Journal of the Faculty of Engineering， 2021， 36（2）： 421-432.
[17]	Aydin H， Bayram A， Esme U， et al. Application of grey relation analysis （GRA） and Taguchi method for the parametric optimization of friction stir welding （FSW） process［J］. Materials and Technology， 2010， 44（4）： 205-211.
[18]	柴宝堂，雷洪，徐猛，等. 基于BP神经网络的RH精炼终点钢液温度预测［J］. 炼钢， 2023， 39（5）： 33-40，47.
	Chai Bao-tang， Lei Hong， Xu Meng， et al. Predicted temperature of molten steel at the end of RH refining on the base of BP neural network［J］. Steelmaking， 2023， 39（5）： 33-40，47.
[19]	Gaïffas S， Merad I， Yu Y Y. WildWood： a new random forest algorithm［J］. IEEE Transactions on Information Theory， 2023， 69（10）： 6586-6604.
[20]	Tarchoune I， Djebba A， Merouani H F， et al. An improved random forest based on feature selection and feature weighting for case retrieval in CBR system application to medical data［J］. International Journal of Software Innovation， 2022， 10（1）： 14-16.
[21]	Gewers F L， Ferreira G R， Arruda H F， et al. Principal component analysis： a natural approach to data exploration［J］. ACM Computing Surveys， 2021， 54（4）： 1-34.
[22]	Izonin I， Tkachenko R， Shakhovska N， et al. A two-step data normalization approach for improving classification accuracy in the medical diagnosis domain［J］. Mathematics， 2022， 10（11）： 1942-1947.
[23]	Li Z H， Qin L， Guo B S， et al. Characterization of the convoluted 3D intermetallic phases in a recycled Al alloy by synchrotron X-ray tomography and machine learning［J］. Acta Metallugica Sinica（English Letters）， 2022， 35（1）： 115-123.
[24]	Zheng Q， Feng B W， Liu Z Y， et al. Application of improved particle swarm optimisation algorithm in hull form optimisation［J］. Journal of Marine Science and Engineering， 2021， 9（9）： 955-962.

特征变量	最大值	最小值	上边缘值	下边缘值	处理前极差	处理后极差
A/℃	1 676	0	1 642	1 593	1 676	49
B×10⁶	872	1	688	243	871	445
C/t	3 882	350	411	365	3 532	46
D×10⁶	830	0	605	212	830	393
E	94	2	86	2	92	84
F/min	77	27	57	34	50	23
G/min	39	13	34	24	26	10
H/℃	1 285	0	1 228	887	1 285	341
I/m³	343	0	204	0	343	204
J/kg	10 186	0	248	0	10 186	248
K/kg	20 669	0	701	638	20 669	63
L/kg	497	168	370	177	329	193
M×10⁶	446	117	405	244	329	161
N×10⁶	1 700	6	27	7	1 694	20

特征变量	最大值	最小值	上边缘值	下边缘值	处理前极差	处理后极差
A/℃	1 676	0	1 642	1 593	1 676	49
B×10⁶	872	1	688	243	871	445
C/t	3 882	350	411	365	3 532	46
D×10⁶	830	0	605	212	830	393
E	94	2	86	2	92	84
F/min	77	27	57	34	50	23
G/min	39	13	34	24	26	10
H/℃	1 285	0	1 228	887	1 285	341
I/m³	343	0	204	0	343	204
J/kg	10 186	0	248	0	10 186	248
K/kg	20 669	0	701	638	20 669	63
L/kg	497	168	370	177	329	193
M×10⁶	446	117	405	244	329	161
N×10⁶	1 700	6	27	7	1 694	20

目标序列	比较序列	灰色关联度
N×10⁶	A/℃	0.841 2
	B×10⁶	0.833 6
	D×10⁶	0.827 1
	I/m³	0.815 3
	E	0.813 0
	F/min	0.800 3
	G/min	0.783 1
	H/℃	0.774 6
	J/kg	0.762 4
	M×10⁶	0.731 3
	K/kg	0.656 8
	L/kg	0.604 5
	C/t	0.530 1

目标序列	比较序列	灰色关联度
N×10⁶	A/℃	0.841 2
	B×10⁶	0.833 6
	D×10⁶	0.827 1
	I/m³	0.815 3
	E	0.813 0
	F/min	0.800 3
	G/min	0.783 1
	H/℃	0.774 6
	J/kg	0.762 4
	M×10⁶	0.731 3
	K/kg	0.656 8
	L/kg	0.604 5
	C/t	0.530 1

排序	特征变量	袋外数据误差
1	K/kg	0.215 3
2	C/t	0.202 6
3	F/min	0.193 5
4	G/min	0.186 2
5	I/m³	0.158 9
6	A/℃	0.125 3
7	J/kg	0.118 2
8	B×10⁶	0.115 8
9	L/kg	0.108 7
10	D×10⁶	0.106 3
11	E	0.104 0
12	H/℃	0.096 9
13	M×10⁶	0.080 4