Prediction of blast furnace parameters using feature engineering and Stacking algorithm

Abstract

Based on the unique characteristics of data within the BF ironmaking domain, this paper select hearth activity, [Si + Ti], and permeability index (PI) as target parameters to verify the effectiveness of the combination of feature engineering and Stacking algorithm in the field of BF process parameter prediction. Based on the actual production data stored in the enterprise database, this paper takes the actual production problems in the process of BF ironmaking as the application background. Through the combination of feature selection and ironmaking theory, the characteristic variables of the prediction model are selected for the preprocessed BF production data, and the accurate prediction of different machine learning algorithms is realized. The results show that the accuracy of stacking algorithm for classification and regression is more than 90%. The model process has good learning and generalization ability to effectively utilize BF ironmaking data and accurately predict BF process parameters.

Get full access to this article

View all access options for this article.

References

Arroba

Risco-Martín

Zapater

et al. Enhancing regression models for complex systems using evolutionary techniques for feature engineering. J Grid Comput. 2015;13(3):409–423. doi:10.1007/s10723-014-9313-8.

Cheng

Wang

et al. Feature selection: a data perspective. ACM Comput Surv. 2017;50(6):1–45. doi:10.1145/3136625.

Thomas

MC.

The best two independent measurements are not the two best. IEEE T Syst Man Cy-S. 1974;4(1):1–16. doi:10.1109/TSMC.1974.5408535.

Liu

Toward integrating feature selection algorithms for classification and clustering. IEEE T Knowl Data En. 2005;17(4):491–492.

Tang

Alelyani

Liu

Feature selection for classification: A Review. Data Classi: Algorithms App. 2014;2014:37–64. doi:10.1201/b17320-3.

Kwak

Choi

CH.

Input feature selection by mutual information based on Parzen window. IEEE Trans Pattern Anal Mach Intell. 2002;24(12):1667–1. doi:10.1109/TPAMI.2002.1114861.

Lazar

Taminau

Meganck

et al. A survey on filter techniques for feature selection in gene expression microarray analysis. IEEE/ACM Trans Comput Biol Bioinform. 2012;9(4):1106–1108. doi:1106. 10.1109/TCBB.2012.33.

Liu

Yang

et al. Clustering-guided sparse structural learning for unsupervised feature selection. IEEE Trans Knowl Data Eng. 2014;26(9):2138–2150. doi:10.1109/TKDE.2013.65.

Liu

et al. Evaluation and prediction of blast furnace status based on big data platform of ironmaking and data mining. ISIJ Int. 2021;61(1):108–108. doi:10.2355/isijinternational.ISIJINT-2020-249.

10.

Sun

Wang

Hybrid event-, mechanism- and data-driven prediction of blast furnace gas generation. Energy. 2020;199:117497–8. doi:10.1016/j.energy.2020.117497.

11.

Zhou

Song

Wang

et al. Data-driven nonlinear subspace modeling for prediction and control of molten iron quality indices in blast furnace ironmaking. IEEE Trans Control Syst Technol. 2017;25:1–16. doi:10.1109/TCST.2016.2631124.

12.

Tathagata

Prediction of silicon content in blast furnace hot metal using partial least squares (PLS). ISIJ Int 2005;45(12):1943–1945. doi:10.2355/isijinternational.45.1943.

13.

Chen

Wang

Han

HL.

Prediction and control for silicon content in pig iron of blast furnace by integrating artificial neural network with genetic algorithm. Ironmak Steelmak. 2009;37(6):458–453. doi:10.1179/174328109X445769.

14.

Gao

Ling

Rules extraction from fuzzy-based blast furnace SVMs multiclassifier for decision-making. IEEE Trans Fuzzy Syst. 2014;22(3):586–586. doi:10.1109/TFUZZ.2013.2269145.

15.

Zhang

et al. Supply and demand forecasting of blast furnace gas based on artificial neural network in iron and steel works. Adv Mater Res. 2012;443:183–188. doi:10.4028/www.scientific.net/AMR.443-444.183.

16.

Zhang

Wang

Improved wavelet neural network to predict blast furnace gas production in iron and steel enterprises. Proceedings of the 2017 International Conference on Computer Graphics and Digital Image Processing (CGDIP 17), Prague. 2017, p. 1–5. doi:10.1145/3110224.3123241.