Enhanced semi-supervised learning for top gas flow state classification to optimize emission and production in blast ironmaking furnaces

Abstract

Automated classification of gas flow states in blast furnaces using top-camera imagery typically demands a large volume of labeled data, whose manual annotation is both labor-intensive and cost-prohibitive. To mitigate this challenge, we present an enhanced semi-supervised learning approach based on the Mean Teacher framework, incorporating a novel feature loss module to maximize classification performance with limited labeled samples. The model studies show that the proposed model surpasses both the baseline Mean Teacher model and fully supervised method in accuracy. Specifically, for datasets with 20%, 30%, and 40% label ratios, using a single training iteration, the model yields accuracies of 78.61%, 82.21%, and 85.2%, respectively, while multiple-cycle training iterations achieves 82.09%, 81.97%, and 81.59%, respectively. Furthermore, scenario-specific training schemes are introduced to support diverse deployment need. These findings highlight the potential of the proposed technique in minimizing labeling requirements and advancing intelligent blast furnace diagnostics.

Keywords

blast furnace / gas flow state / semi-supervised learning / mean teacher / feature loss

References

[1] Zhang SR, Pan GY, Liu M. Development route of China’s ironmaking technology under the “dual carbon” goal. Ironmaking, 2022, 41(6): 1

[2] Zuo HB, Guo LF, Wang YJ, Zheng J. Effect of bosh angle and shaft angle on gas distribution in blast furnace. Iron Steel, 2018, 53(2): 20

[3] W.B. Wu, S.B. Kuang, L.L. Jiao, and A.B. Yu, A machine learning model for quickly predicting the inner states of iron-making blast furnaces, Powder Technol., 445(2024), art. No. 120137.

[4] S.B. Kuang, Z.Y. Li, and A.B. Yu, Review on modeling and simulation of blast furnace, Steel Res. Int., 89(2018), No. 1, art. No. 1700071.

[5] Zhao ZJ, She XF, Zhao YZ, et al. . Numerical simulation study on the influence of burden profile on gas distribution in blast furnace. J. Mater. Metall., 2023, 22(1): 38

[6] Xu YJ, She XF, Li HB, Cao WQ, Zhang JB, Xue QG. Simulation of gas flow distribution in blast furnace under different column modes. China Metall., 2023, 33(3): 54

[7] Z.Y. Li, Z. Qi, Z.X. Wang, L. Ji, and Z. Li, Numerical investigation of scrap charging into an ironmaking blast furnace, Energy, 315(2025), art. No. 134440.

[8] Jiao LL, Zhang XY, Kuang SB, Yu AB. Numerical investigation of particle size on the performance of ironmaking blast furnace. Metall. Mater. Trans. B, 2024, 55(5): 3387.

[9] J. Li, S.B. Kuang, L.L. Jiao, L.L. Liu, R.P. Zou, and A.B. Yu, Numerical modeling and analysis of hydrogen blast furnace ironmaking process, Fuel, 323(2022), art. No. 124368.

[10] Bambauer F, Wirtz S, Scherer V, Bartusch H. Transient DEM–CFD simulation of solid and fluid flow in a three dimensional blast furnace model. Powder Technol., 2018, 334: 53.

[11] Hou QF, E DY, Kuang SB, Li ZY, Yu AB. DEM-based virtual experimental blast furnace: A quasi-steady state model. Powder Technol., 2017, 314: 557.

[12] R. Roeplal, Y.S. Pang, A. Adema, J. van der Stel, and D. Schott, Modelling of phenomena affecting blast furnace burden permeability using the Discrete Element Method (DEM)–A review, Powder Technol., 415(2023), art. No. 118161.

[13] Xin ZC, Zhang JS, Peng KX, et al. . Explainable machine learning model for predicting molten steel temperature in the LF refining process. Int. J. Miner. Metall. Mater., 2024, 31(12): 2657.

[14] Wang MY, Tang J, Chu MS, Shi Q, Zhang Z. Prediction and optimization of flue pressure in sintering process based on SHAP. Int. J. Miner. Metall. Mater., 2025, 32(2): 346.

[15] H.Y. Jiang, G.Q. Xu, and J.C. Zhou, Application of bp algorithm in the pattern recognition of gas fluid distribution in blast furnace, J. Northeast. Univ. Nat. Sci., (2000), No. 2, p. 144.

[16] Cui GM, Cao AW, Ma X. An identification method of center gas-flow distribution pattern based on sensed infrared image processing. Inf. Control, 2014, 43(1): 110

[17] An JQ, Wu M, He Y, Cao WH. Soft-sensing method of gas flow distribution of blast furnace burden surface based on multi-level hierarchical fusion algorithm. Acta Autom. Sin., 2011, 37(4): 496.

[18] C.S. Xue, W.H. Cao, M. Wu, and Y. He, Recognition method for determining gas flow distribution along blast furnace burden surface, J. Tsinghua Univ. Sci. Technol., (2008), No. S2, p. 1785.

[19] Shi L, Wen YB, Zhao GS, Yu T. Recognition of blast furnace gas flow center distribution based on infrared image processing. J. Iron Steel Res. Int., 2016, 23(3): 203.

[20] J.J. Wang and L. Shi, Classification of blast furnace gas flow distribution states based on singular energy values, Inf. Technol. Informatization, (2022), No. 9, p. 39.

[21] Dang XJ, Shi L, Zhao N, Yuan DF, Li JP. Prediction of utilization ratio of blast furnace gas based on parameter optimized by SVR method. J. Iron Steel Res., 2021, 33(4): 279

[22] Han B, Shi L, Li MX, Yu T. Research on intelligent processing technology of infrared video image of blast furnace top. J. Inn. Mongolia Univ. Sci. Technol., 2019, 38(1): 8

[23] Shi L, Liu WL, Li JP, Zhao N. Prediction model of gas utilization rate based on blast furnace throat gas distribution. J. Inn. Mongolia Univ. Sci. Technol., 2016, 35(2): 121

[24] Wu M, Wang CJ, An JQ, He Y. A recognition method for gas flow distribution based on temperature field of burden surface in blast furnace. Inf. Control, 2011, 40(1): 78

[25] Zhang SH, Shi L, Li MX, Yu T. Calibration of temperature field of gas flow in different development stages based on big data. J. Inn. Mongolia Univ. Sci. Technol., 2018, 37(4): 343

[26] Derry A, Krzywinski M, Altman N. Convolutional neural networks. Nat. Methods, 2023, 20(9): 1269.

[27] He Y, Xiao LG. Structured pruning for deep convolutional neural networks: A survey. IEEE Trans. Pattern Anal. Mach. Intell., 2024, 46(5): 2900.

[28] Li ZW, Liu F, Yang WJ, Peng SH, Zhou J. A survey of convolutional neural networks: Analysis, applications, and prospects. IEEE Trans. Neural Netw. Learn. Syst., 2022, 33(12): 6999.

[29] Liu R, Zhao WG, Liu S, et al. . Intelligent prediction and realtime monitoring system for gas flow distribution at the top of blast furnace. ISIJ Int., 2023, 63(10): 1714.

[30] Zhang TF, Zhang XL, Han T, Shi ZJ, Guo YQ, Wang HY. Application of artificial intelligence image recognition technology in blast furnace tuyere monitoring. Metall. Ind. Autom., 2021, 45(3): 58

[31] R.H. Wang, Z.Y. Li, L.Z. Yang, et al., Application of efficient channel attention residual mechanism in blast furnace tuyere image anomaly detection, Appl. Sci., 12(2022), No. 15, art. No. 7823.

[32] Zhao ZW, Li QQ, Liu S, et al. . Research on blast furnace air outlet state identification model based on improved ResNet18. Arabian J. Sci. Eng., 2025, 50(10): 6943.

[33] Li FM, Wang J, Liu XJ, Duan YF, Zhang XS, Lyu Q. Multi-scale extraction and recognition model of tuyere image based on CNN-GraphSAGE. Iron Steel, 2025, 60(1): 40.

[34] Yang XL, Song ZX, King I, Xu ZL. A survey on deep semi-supervised learning. IEEE Trans. Knowl. Data Eng., 2023, 35(9): 8934.

[35] Creswell A, White T, Dumoulin V, Arulkumaran K, Sengupta B, Bharath AA. Generative adversarial networks: An overview. IEEE Signal Process. Mag., 2018, 35(1): 53.

[36] Goodfellow I, Pouget-Abadie J, Mirza M, et al. . Generative adversarial networks. Commun. ACM, 2020, 63(11): 139.

[37] Jing LL, Tian YL. Self-supervised visual feature learning with deep neural networks: A survey. IEEE Trans. Pattern Anal. Mach. Intell., 2021, 43(11): 4037.

[38] Liu X, Zhang FJ, Hou ZY, et al. . Self-supervised learning: Generative or contrastive. IEEE Trans. Knowl. Data Eng., 2023, 35(1): 857

[39] J.M. Duarte and L. Berton, A review of semi-supervised learning for text classification, Artif. Intell. Rev., (2023), p. 1.

[40] Han M, Wu HX, Chen ZQ, Li MH, Zhang XL. A survey of multi-label classification based on supervised and semisupervised learning. Int. J. Mach. Learn. Cybern., 2023, 14(3): 697.

[41] Zhou SG, Xue ZH, Du PJ. Semisupervised stacked autoencoder with cotraining for hyperspectral image classification. IEEE Trans. Geosci. Remote Sens., 2019, 57(6): 3813.

[42] X.T. Zheng, H.W. Cui, C.J. Xu, and X.Q. Lu, Dual teacher: A semisupervised cotraining framework for cross-domain ship detection, IEEE Trans. Geosci. Remote Sens., 61(2023), art. No. 5613312.

[43] B.T. Li, D.C. Pi, and Y.X. Lin, Learning ladder neural networks for semi-supervised node classification in social network, Expert Syst. Appl., 165(2021), art. No. 113957.

[44] P. Meel and D.K. Vishwakarma, A temporal ensembling based semi-supervised ConvNet for the detection of fake news articles, Expert Syst. Appl., 177(2021), art. No. 115002.

[45] Tian Y, Zhang LG, Sun JG, Yin GS, Dong YX. Consistency regularization teacher–student semi-supervised learning method for target recognition in SAR images. Visual. Comput., 2022, 38(12): 4179.

[46] Z. Xu, Y.X. Wang, D.H. Lu, et al., Ambiguity-selective consistency regularization for mean-teacher semi-supervised medical image segmentation, Med. Image Anal., 88(2023), art. No. 102880.

[47] Wei J, Ma Q, Wang WT, Guo H, Liu Z, Zhang JJ. Abnormal area identification of corn ear based on semi-supervised learning. IET Image Process., 2022, 16(9): 2351.

[48] G. Turk, M. Ozdemir, R. Zeydan, Y. Turk, Z. Bilgin, and E. Zeydan, On the identification of thyroid nodules using semi-supervised deep learning, Int. J. Numer. Methods Biomed. Eng., 37(2021), No. 3, art. No. e3433.

[49] S. Shim, J. Kim, S.W. Lee, and G.C. Cho, Road damage detection using super-resolution and semi-supervised learning with generative adversarial network, Autom. Constr., 135(2022), art. No. 104139.

[50] Guo JJ, Wang Q, Li YT. Semi-supervised learning based on convolutional neural network and uncertainty filter for façade defects classification. Comput. Aided Civ. Infrastruct. Eng., 2021, 36(3): 302.

[51] Yan ZJ, Wang Y, Chen Y, Zhang C. Semi-supervised rock slice image classification based on hierarchy consistency mean teacher model. J. Appl. Sci., 2024, 42(1): 27

[52] Fan Y, Kukleva A, Dai DX, Schiele B. Revisiting consistency regularization for semi-supervised learning. Int. J. Comput. Vis., 2023, 131(3): 626.

[53] E. Koh, Y.J. Lee, and S.B. Kim, GraphixMatch: Improving semi-supervised learning for graph classification with FixMatch, Neurocomputing, 607(2024), art. No. 128356.

[54] S. Aboobacker, D. Vijayasenan, S. Sumam David, P.K. Suresh, and S. Sreeram, Semantic segmentation of low magnification effusion cytology images: A semi-supervised approach, Comput. Biol. Med., 150(2022), art. No. 106179.

RIGHTS & PERMISSIONS

University of Science and Technology Beijing