Industrial big data analysis strategy based on automatic data classification and interpretable knowledge graph

Abstract

Machine learning has emerged as a critical tool for processing the complex and large-scale datasets generated in the steel industry. However, a single machine learning model struggles to capture all relevant information owing to the variety of steel grades, thereby limiting its extensibility and broader industrial application. Furthermore, most machine-learning models are “black boxes” with low interpretability. Therefore, this paper proposes a novel strategy for industrial big data analysis. First, a data classification model was developed using unsupervised clustering techniques to automatically divide the dataset into four distinct classes. Simultaneously, key physical metallurgy (PM) variables were calculated and incorporated as input features to improve property prediction. Next, an interpretable knowledge graph was constructed for each class, connecting the relevant features with the PM variables. Using these graphs, a graph convolutional network (GCN) model was developed for each class to predict the steel properties. The results demonstrate that this approach delivers better predictions than models without automatic data classification. Furthermore, compared to traditional deep learning models, GCN models based on interpretable knowledge graphs provide superior prediction accuracy and significantly improved interpretability and extensibility.

Keywords

Industrial big data / property prediction / data classification / physical metallurgy / graph convolutional network

References

[1] Zhang Q.Research on mechanical property prediction of hot rolled steel based on lightweight multi-branch convolutional neural network.Mater Today Commun2023;37:107445

[2] Liu S,Zhang S.Study on the prediction of tensile strength and phase transition for ultra-high strength hot stamping steel.J Mater Res Technol2020;9:14244-53

[3] dos Santos, A. A.; Barbosa, R. Model for microstructure prediction in hot strip rolled steels.Steel Res Int2010;81:55-63

[4] Han HN,Kim HJ.A model for deformation, temperature and phase transformation behavior of steels on run-out table in hot strip mill.J Mater Process Technol2002;128:216-25

[5] Zhang X,Tieu A,Wang G.Numerical modelling of the thermal deformation of CVC roll in hot strip rolling.J Mater Process Technol2002;130-1:219-23

[6] Militzer M,Meadowcroft TR.Microstructural model for hot strip rolling of high-strength low-alloy steels.Metall Mater Trans A2000;31:1247-59

[7] Wei X,Jia Z,Xu W.On the use of transfer modeling to design new steels with excellent rotating bending fatigue resistance even in the case of very small calibration datasets.Acta Mater2022;235:118103

[8] Lee J,Lee Y.Design of high strength medium-Mn steel using machine learning.Mater Sci Eng A2022;843:143148

[9] Abd-Elaziem W,Sebaey TA,Abd El-baky MA.Machine learning for advancing laser powder bed fusion of stainless steel.J Mater Res Technol2024;30:4986-5016

[10] Geng X,Wu H.Data-driven and artificial intelligence accelerated steel material research and intelligent manufacturing technology.MGE Advances2023;1:e10

[11] Zhu L,Chen Q.Prediction of ultimate tensile strength of Al-Si alloys based on multimodal fusion learning.MGE Advances2024;2:e26

[12] Shi Z,He X.Prediction model of yield strength of V–N steel hot-rolled plate based on machine learning algorithm.JOM2023;75:1750-62

[13] Xu G,Lü Z,Xu J.Prediction of mechanical properties for deep drawing steel by deep learning.Int J Miner Metall Mater2023;30:156-65

[14] He X,Tian T.Prediction of mechanical properties of hot rolled strips with generalized RBFNN and composite expectile regression.IEEE Access2022;10:106534-42

[15] Xu Z,Zhang K.Mechanical properties prediction for hot rolled alloy steel using convolutional neural network.IEEE Access2019;7:47068-78

[16] Jiang X,Zhang G.A strategy combining machine learning and multiscale calculation to predict tensile strength for pearlitic steel wires with industrial data.Scr Mater2020;186:272-7

[17] Guo S,Liu X,Jiang Q.A predicting model for properties of steel using the industrial big data based on machine learning.Comput Mater Sci2019;160:95-104

[18] Li W,Zhao Y,Wang W.Prediction model for mechanical properties of hot-rolled strips by deep learning.J Iron Steel Res Int2020;27:1045-53

[19] Xie Q,Li J,Cai J.Online prediction of mechanical properties of hot rolled steel plate using machine learning.Mater Design2021;197:109201

[20] Yang Z,Xu F.Online prediction of mechanical properties of the hot rolled steel plate using time-series deep neural network.ISIJ Int2023;63:746-57

[21] Cui C,Li X,Liu J.A strategy combining machine learning and physical metallurgical principles to predict mechanical properties for hot rolled Ti micro-alloyed steels.J Mater Process Technol2023;311:117810

[22] Li F,Song Y.Deep learning for predictive mechanical properties of hot-rolled strip in complex manufacturing systems.Int J Miner Metall Mater2023;30:1093-103

[23] Li H,Huang J.Physical metallurgy guided industrial big data analysis system with data classification and property prediction.Steel Res Int2022;93:2100820

[24] Thomas A,Alam M,Sack H.Materials fatigue prediction using graph neural networks on microstructure representations.Sci Rep2023;13:12562 PMCID:PMC10397301

[25] Dai M,Liang Y.Graph neural networks for an accurate and interpretable prediction of the properties of polycrystalline materials.npj Comput Mater2021;7:574

[26] Sadeghpour E.Data-driven models for structure-property prediction in additively manufactured steels.Comput Mater Sci2022;215:111782

[27] Karimi K,Mulewska K.Prediction of steel nanohardness by using graph neural networks on surface polycrystallinity maps.Scr Mater2023;234:115559

[28] Li Y,Zhang Y.Thermodynamically informed graph for interpretable and extensible machine learning: martensite start temperature prediction.Calphad2024;85:102710

[29] Shi X,Huang Y,Hong Z.A review on the applications of graph neural networks in materials science at the atomic scale.MGE Advances2024;2:e50

[30] Celebi ME,Vela PA.A comparative study of efficient initialization methods for the k-means clustering algorithm.Expert Syst Appl2013;40:200-10

[31] Liu Y,Wang Z.Predicting creep rupture life of Ni-based single crystal superalloys using divide-and-conquer approach based machine learning.Acta Mater2020;195:454-67

[32] Pourbahrami S,Khanli LM.A survey of neighborhood construction algorithms for clustering and classifying data points.Comput Sci Rev2020;38:100315

[33] Hussain SF.A k-means based co-clustering (kCC) algorithm for sparse, high dimensional data.Expert Syst Appl2019;118:20-34

[34] Tzortzis G.The MinMax k-Means clustering algorithm.Pattern Recogn2014;47:2505-16

[35] Tan X,Rao X.Effect of ultra-fast heating on microstructure and mechanical properties of cold-rolled low-carbon low-alloy Q&P steels with different austenitizing temperature.Mater Charact2022;191:112086