A feature selection method for desulfurizer addition prediction based on importance measure

ZHAO Haijie; DAN Binbin; LIU Yang; REN Zeyu; DU Liping; ZHOU Chun

doi:10.7513/j.issn.1004-7638.2025.05.005

Volume 46 Issue 5

Oct. 2025

Turn off MathJax

Article Contents

Article Navigation > IRON STEEL VANADIUM TITANIUM > 2025 > 46(5): 46-53, 64

ZHAO Haijie, DAN Binbin, LIU Yang, REN Zeyu, DU Liping, ZHOU Chun. A feature selection method for desulfurizer addition prediction based on importance measure[J]. IRON STEEL VANADIUM TITANIUM, 2025, 46(5): 46-53, 64. doi: 10.7513/j.issn.1004-7638.2025.05.005

Citation:

ZHAO Haijie, DAN Binbin, LIU Yang, REN Zeyu, DU Liping, ZHOU Chun. A feature selection method for desulfurizer addition prediction based on importance measure[J]. IRON STEEL VANADIUM TITANIUM, 2025, 46(5): 46-53, 64. doi: 10.7513/j.issn.1004-7638.2025.05.005

Citation:

PDF( 1677 KB)

A feature selection method for desulfurizer addition prediction based on importance measure

doi: 10.7513/j.issn.1004-7638.2025.05.005

ZHAO Haijie^{1, 2
,},
DAN Binbin^{1, 2
,
,},
LIU Yang³,
REN Zeyu⁴,
DU Liping^{1, 2},
ZHOU Chun^{1, 2}

1.
Key Laboratory of Metallurgical Equipment and Control Technology, Wuhan University of Science and Technology, Ministry of Education, Wuhan 430081, Hubei, China
2.
Hubei Key Laboratory of Mechanical Transmission and Manufacturing Engineering, Wuhan University of Science and Technology, Wuhan 430081, Hubei, China
3.
Baosteel Central Research Institute (Wuhan Iron and Steel Limited Technology Centre), Wuhan 430080, Hubei, China
4.
School of Physics and Electronic Science, Hunan Institute of Science and Technology, Yueyang 414006, Hubei, China

More Information

Received Date: 2025-07-22
Accepted Date: 2025-09-05
Rev Recd Date: 2025-09-05
Publish Date: 2025-10-30

Abstract

Abstract

Aiming at the problems of high parameter dimension, strong feature redundancy and weak correlation between target variables and features in hot metal KR desulfurization production process, an integrated feature selection method IMFS (Feature selection based on importance measure) based on importance measure is proposed. In the filtering pre-screening stage, the maximal mutual information coefficient ( MIC ) is used to measure the correlation between each parameter and the target variable, as well as the redundancy among each parameter, and the scale of candidate parameters is reduced according to the maximum relevance and minimal redundancy criteria. In the embedded selection stage, the LightGBM algorithm is introduced as the supporting model for quantifying information contribution and data sensitivity, and the entropy weight method is used to weight and fuse the dual measurement results. Finally, according to the feature importance coefficient, the feature subset is optimized by combining the GBT sequential forward search strategy. The experimental results show that compared with other methods, IMFS has significant advantages in eliminating redundant features and improving prediction accuracy, and can effectively balance the number of features and prediction accuracy.
- desulfurizer addition,
- feature selection,
- importance coefficient,
- double metric,
- search strategy

FullText(HTML)

References(13)

References

[1]	GAO J, CUI L, WANG W, et al. Prediction of sulfur content during steel refining process based on machine learning methods[J]. Steel Research International, 2024, 96(3): 2400662-2400662.
[2]	GONG H J, LIANG X T, ZHOU Z C, et al. Application of rotary injection desulfurization technology in hot metal pretreatment[J]. Iron Steel Vanadium Titanium, 2020, 41(1): 173-178. (龚洪君, 梁新腾, 周遵传, 等. 旋转喷吹脱硫技术在铁水预处理上的应用研究[J]. 钢铁钒钛, 2020, 41(1): 173-178. doi: 10.7513/j.issn.1004-7638.2020.01.030 GONG H J, LIANG X T, ZHOU Z C, et al. Application of rotary injection desulfurization technology in hot metal pretreatment[J]. Iron Steel Vanadium Titanium, 2020, 41（1）: 173-178. doi: 10.7513/j.issn.1004-7638.2020.01.030
[3]	ADHIWIGUNA IBGS, KARAGÜLMEZ G, KESKIN O, et al. Investigation on applicability of lime as desulfurization agent for molten cast iron[J]. Steel Research International, 2025, 96(1): 2400416.
[4]	ZHENG Y, ZUO K L. Prediction model of desulfurizer consumption based on BP neural network and regression[J]. Iron Steel Vanadium Titanium, 2017, 38(4): 130-134. (郑毅, 左康林. 基于BP神经网络和回归的脱硫粉剂预报模型[J]. 钢铁钒钛, 2017, 38(4): 130-134. doi: 10.7513/j.issn.1004-7638.2017.04.023 ZHENG Y, ZUO K L. Prediction model of desulfurizer consumption based on BP neural network and regression[J]. Iron Steel Vanadium Titanium, 2017, 38（4）: 130-134. doi: 10.7513/j.issn.1004-7638.2017.04.023
[5]	LIU Z X, DU J Q, LUO J G, et al. Review on stability feature selection[J]. Computer Engineering and Applications, 2025, 61(7): 81-95. (刘梓萱, 杜建强, 罗计根, 等. 稳定性特征选择研究综述[J]. 计算机工程与应用, 2025, 61(7): 81-95. doi: 10.3778/j.issn.1002-8331.2406-0410 LIU Z X, DU J Q, LUO J G, et al. Review on stability feature selection[J]. Computer Engineering and Applications, 2025, 61（7）: 81-95. doi: 10.3778/j.issn.1002-8331.2406-0410
[6]	WANG N, LI X F, NIE L D, et al. High-precision vehicle energy consumption prediction using mutual information feature selection[J]. Journal of Tongji University (Natural Science), 2024, 52(S1): 39-45. (王宁, 李秀峰, 聂辽栋, 等. 基于MI特征选择的车辆能耗高精度预测方法[J]. 同济大学学报(自然科学版), 2024, 52(S1): 39-45. doi: 10.11908/j.issn.0253-374x.24794 WANG N, LI X F, NIE L D, et al. High-precision vehicle energy consumption prediction using mutual information feature selection[J]. Journal of Tongji University (Natural Science), 2024, 52（S1）: 39-45. doi: 10.11908/j.issn.0253-374x.24794
[7]	YAN X M, CHEN C, WANG N, et al. Prediction of desulfurization rate during LF refining process based on random search and AdaBoost model[J]. Journal of Materials and Metallurgy, 2023, 22(5): 430-436, 443. (严旭梅, 陈超, 王楠, 等. 基于随机搜索算法和AdaBoost模型预测LF精炼过程脱硫率[J]. 材料与冶金学报, 2023, 22(5): 430-436, 443. YAN X M, CHEN C, WANG N, et al. Prediction of desulfurization rate during LF refining process based on random search and AdaBoost model[J]. Journal of Materials and Metallurgy, 2023, 22（5）: 430-436, 443.
[8]	FANG Y F, DAN B B, WU J W, et al. Method for predicting desulfurizer dosage based on ensemble learning[J]. Journal of Wuhan University of Science and Technology, 2024, 47(5): 361-367. (方一飞, 但斌斌, 吴经纬, 等. 基于集成学习的脱硫剂加入量预测方法[J]. 武汉科技大学学报, 2024, 47(5): 361-367. doi: 10.3969/j.issn.1674-3644.2024.05.006 FANG Y F, DAN B B, WU J W, et al. Method for predicting desulfurizer dosage based on ensemble learning[J]. Journal of Wuhan University of Science and Technology, 2024, 47（5）: 361-367. doi: 10.3969/j.issn.1674-3644.2024.05.006
[9]	XU M, LEI H, HE J Y, et al. Predicting the endpoint steel temperature of RH refining using improved XGBoost[J]. Journal of Materials and Metallurgy, 2023, 22(5): 437-443. (徐猛, 雷洪, 何江一, 等. 利用改进XGBoost预测RH精炼终点钢水温度[J]. 材料与冶金学报, 2023, 22(5): 437-443. XU M, LEI H, HE J Y, et al. Predicting the endpoint steel temperature of RH refining using improved XGBoost[J]. Journal of Materials and Metallurgy, 2023, 22（5）: 437-443.
[10]	GU T Y, GUO J S, LI Z X, et al. Detecting associations based on the multi-variable maximum information coefficient[J]. IEEE Access, 2021, 9: 54912-54922. doi: 10.1109/ACCESS.2021.3070925
[11]	JU Y, SUN G Y, CHEN Q H, et al. A model combining convolutional neural network and LightGBM algorithm for ultra-short-term wind power forecasting[J]. IEEE Access, 2019: 28309-28318.
[12]	LI Y Z, DAI W, ZHANG W F. Bearing fault feature selection method based on weighted multidimensional feature fusion[J]. IEEE Access, 2020, 8: 19008-19025. doi: 10.1109/ACCESS.2020.2967537
[13]	ZHANG S G, ZHOU T, SUN L, et al. v-Support vector regression model based on Gauss-Laplace mixture noise characteristic for wind speed prediction[J]. Entropy, 2019, 21(11): 1056.