Overview: Integration and Development of Physical Models and Artificial Intelligence in Alloy Design
WANG Chenchong, XU Wei()
State Key Laboratory of Digital Steel, Northeastern University, Shenyang 110819, China
Cite this article:
WANG Chenchong, XU Wei. Overview: Integration and Development of Physical Models and Artificial Intelligence in Alloy Design. Acta Metall Sin, 2025, 61(4): 541-560.
With the rise of data-driven methods as the fourth scientific paradigm, their impact on the third paradigm—physical model-driven approaches—has been significant in the field of alloy design. However, neither paradigm can overcome the trade-off between model accuracy and interpretability, particularly in mechanical performance design. As a consequence, they fail to meet the efficiency and rationality requirements necessary for alloy development within Material Genome Engineering, especially for metal structural materials. This challenge has led to the emergence of the fifth paradigm, AI4Sci, in alloy design. This article provides an overview of various cases employing the physical metallurgy-guided artificial intelligence method system. It systematically explains how to integrate physical models and mechanisms with artificial intelligence at three levels: numerical data guidance, image data guidance, and mechanism guidance. This approach aims to resolve the inherent trade-off between accuracy and interpretability in alloy design. In addition, it explores the theoretical foundations, advantages, and limitations of three paradigms—multi-scale physical models, artificial intelligence, and AI4Sci—within the field. For cross-scale modeling and materials science large models, this article offers insights into conceptual frameworks and technical methodologies for the future development of each scientific paradigm in alloy design.
Fig.1 Essential nature of the mutually exclusive relationship between model interpretability and accuracy (a) trade-off between model accuracy and interpretability (AI—artificial intelligence) (b) high degrees of freedom in equations (c) low degrees of freedom in equations (d) high degrees of freedom in equations constrained by physics
Fig.2 Quantitative effect of dimensionality reduction methods on model accuracy[29-32] (PCA—principal components analysis, KPCA—kernel based principal component analysis, t-SNE—t-distributed stochastic neighbor embedding, ISOMAP—Isometric Mapping)
Fig.3 Cross scale system of physical metallurgy-guided artificial intelligence (ML—machine learning, Vm—mole fraction, UTS—ultimate tensile strength, UE—uniform elongation, BC—band contrast, KAM—kernel average misorientation, CAM—Class Activation Mapping)
Fig.4 Comparisons of accuracy and extensibility of martensite transformation start temperature (Ms) computation models[43] (MAE—mean absolute error, SVM—support vector machine, DDM-CNN—deep data mining guided convolutional neural network)
Fig.5 Paradigm of thermodynamic mechanism information-guided artificial intelligence for alloy design[10,46-54] (ANN—artificial neural network)
Fig.6 Paradigm of multi-scale mechanism information-guided artificial intelligence for design[60-67] (DFT—density functional theory, FEM—finite element model)
Fig.7 Comparisons of algorithms for guiding artificial intelligence with image core information[73-75] (DL—deep learning, SAM—segment anything model, ML1 and ML2 represent different machine learning models for regression) (a1-a12) large sample, low accuracy (b1-b5) small sample, high accuracy (c1-c16) no sample, medium accuracy
Fig.8 Various multimodal algorithm frameworks and their advantages and disadvantages
Fig.9 Mechanism correlation and data requirements of various image algorithms[73,76,87,90]
Fig.10 Methods and applications of mechanism guided artificial intelligence (PINN—physics-informed neural network, NN—neural network, PDE—partial differential equation, MSE—mean squared error. x—spatial variable, t—temporal variable, w—weight, b—bias, σ—activation function, u—solution of the partial differential equation and also the output of the neural network, L—differential operator, g—known function on the right-hand side of the partial differential equation, θ—parameter set of the partial differential equation, R—residual value, ε—threshold. MSE{u, BC, IC}—mean squared error of the solution u when considering boundary conditions (BC) and initial conditions (IC), MSER is the mean squared error based on the R)
Fig.11 Future development directions of the five scientific paradigms in the field of alloy design
1
Liao M Q, Wang Y, Wang Y, et al. Zentropy theory: Bridging materials gene to materials properties [J]. Acta Metall. Sin., 2024, 60: 1379
doi: 10.11900/0412.1961.2024.00147
Xie J X, Su Y J, Xue D Z, et al. Machine learning for materials research and development [J]. Acta Metall. Sin., 2021, 57: 1343
doi: 10.11900/0412.1961.2021.00357
Agrawal A, Choudhary A. Perspective: Materials informatics and big data: Realization of the “fourth paradigm” of science in materials science [J]. APL Mater., 2016, 4: 053208
4
Peng G C Y, Alber M, Tepole A B, et al. Multiscale modeling meets machine learning: What can we learn? [J]. Arch. Comput. Methods Eng., 2021, 28: 1017
5
Su Y J, Fu H D, Bai Y, et al. Progress in materials genome engineering in China [J]. Acta Metall. Sin., 2020, 56: 1313
doi: 10.11900/0412.1961.2020.00199
Zhang X, Wang L M, Helwig J, et al. Artificial intelligence for science in quantum, atomistic, and continuum systems [DB/OL]. arXiv: 2307. 08423, 2023
7
Cui P, Athey S. Stable learning establishes some common ground between causal inference and machine learning [J]. Nat. Mach. Intell., 2022, 4: 110
8
Zhong X T, Gallagher B, Liu S S, et al. Explainable machine learning in materials science [J]. npj Comput. Mater., 2022, 8: 204
9
Shen C G, Wang C C, Rivera-Díaz-del-Castillo P E J, et al. Discovery of marageing steels: Machine learning vs. physical metallurgical modelling [J]. J. Mater. Sci. Technol., 2021, 87: 258
10
Shen C G, Wang C C, Wei X L, et al. Physical metallurgy-guided machine learning and artificial intelligent design of ultrahigh-strength stainless steel [J]. Acta Mater., 2019, 179: 201
11
Guo A X Y, Cheng L J, Zhan S, et al. Biomedical applications of the powder‐based 3D printed titanium alloys: A review [J]. J. Mater. Sci. Technol., 2022, 125: 252
doi: 10.1016/j.jmst.2021.11.084
12
Cisternas L A, Lucay F A, Botero Y L. Trends in modeling, design, and optimization of multiphase systems in minerals processing [J]. Minerals, 2020, 10: 22
13
Huang L, Ruan S, Xing Y C, et al. A review of uncertainty quantification in medical image analysis: Probabilistic and non-probabilistic methods [J]. Med. Image Anal., 2024, 97: 103223
14
Szymanski N J, Rendy B, Fei Y X, et al. An autonomous laboratory for the accelerated synthesis of novel materials [J]. Nature, 2023, 624: 86
15
Xue D Z, Balachandran P V, Hogden J, et al. Accelerated search for materials with targeted properties by adaptive design [J]. Nat. Commun., 2016, 7: 11241
doi: 10.1038/ncomms11241
pmid: 27079901
16
Zhao W C, Zheng C, Xiao B, et al. Composition refinement of 6061 aluminum alloy using active machine learning model based on Bayesian optimization sampling [J]. Acta Metall. Sin., 2021, 57: 797
doi: 10.11900/0412.1961.2020.00298
Gawlikowski J, Tassi C R N, Ali M, et al. A survey of uncertainty in deep neural networks [J]. Artif. Intell. Rev., 2023, 56(suppl.1) : 1513
18
Kurz A, Hauser K, Mehrtens H A, et al. Uncertainty estimation in medical image classification: Systematic review [J]. JMIR Med. Inf., 2022, 10: e36427
19
Jospin L V, Laga H, Boussaid F, et al. Hands-on Bayesian neural networks—A tutorial for deep learning users [J]. IEEE Comput. Intell. Mag., 2022, 17: 29
20
Venkatraman A, de Oca Zapiain D M, Kalidindi S R. Reduced-order models for ranking damage initiation in dual-phase composites using Bayesian neural networks [J]. JOM, 2020, 72: 4359
doi: 10.1007/s11837-020-04387-y
21
Wu S W, Zhou X G, Chen Q Y, et al. Development of constitutive models for extrapolative prediction of Nb-Ti micro alloyed steel [J]. Steel Res. Int., 2017, 88: 1700082
22
Muth A, Venkatraman A, John R, et al. Neighborhood spatial correlations and machine learning classification of fatigue hot-spots in Ti-6Al-4V [J]. Mech. Mater., 2023, 182: 104679
23
He Z C, Huo S L, Li E, et al. Data-driven approach to characterize and optimize properties of carbon fiber non-woven composite materials [J]. Compos. Struct., 2022, 297: 115961
24
Norris C, Ayyaswamy A, Vishnugopi B S, et al. Uncertainty quantification and propagation in lithium-ion battery electrodes using Bayesian convolutional neural networks [J]. Energy Storage Mater., 2024, 67: 103251
25
Ali S, Abuhmed T, El-Sappagh S, et al. Explainable Artificial Intelligence (XAI): What we know and what is left to attain Trustworthy Artificial Intelligence [J]. Inf. Fusion, 2023, 99: 101805
26
Hassija V, Chamola V, Mahapatra A, et al. Interpreting black-box models: A review on explainable artificial intelligence [J]. Cogn. Comput., 2024, 16: 45
27
Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need [A]. 31st International Conference on Neural Information Processing Systems [C]. Long Beach: Curran Associates Inc., 2017
28
Xu P C, Ji X B, Li M J, et al. Small data machine learning in materials science [J]. npj Comput. Mater., 2023, 9: 42
29
Ikram S T, Cherukuri A K. Improving accuracy of intrusion detection model using PCA and optimized SVM [J]. J. Comput. Inf. Technol., 2016, 24: 133
30
He F, Zhang L Y. Prediction model of end-point phosphorus content in BOF steelmaking process based on PCA and BP neural network [J]. J. Process Control, 2018, 66: 51
31
Xue J L, Chen Y C, Li O, et al. Classification and identification of unknown network protocols based on CNN and T-SNE [J]. J. Phys.: Conf. Ser., 2020, 1617: 012071
32
Gallos I K, Gkiatis K, Matsopoulos G K, et al. ISOMAP and machine learning algorithms for the construction of embedded functional connectivity networks of anatomically separated brain regions from resting state fMRI data of patients with Schizophrenia [J]. AIMS Neurosci., 2021, 8: 295
doi: 10.3934/Neuroscience.2021016
pmid: 33709030
33
Jiang L, Fu H D, Zhang H T, et al. Physical mechanism interpretation of polycrystalline metals' yield strength via a data-driven method: A novel Hall-Petch relationship [J]. Acta Mater., 2022, 231: 117868
34
Baskes M I. The status role of modeling and simulation in materials science and engineering [J]. Curr. Opin. Solid State Mater. Sci., 1999, 4: 273
35
Elliott J A. Novel approaches to multiscale modelling in materials science [J]. Int. Mater. Rev., 2011, 56: 207
36
Chen Z, Haykin S. On different facets of regularization theory [J]. Neural Comput., 2002, 14: 2791
pmid: 12487794
37
Lu Q, Xu W, van der Zwaag S. The design of a compositionally robust martensitic creep-resistant steel with an optimized combination of precipitation hardening and solid-solution strengthening for high-temperature use [J]. Acta Mater., 2014, 77: 310
38
Wang C C, Wei X L, Ren D, et al. High-throughput map design of creep life in low-alloy steels by integrating machine learning with a genetic algorithm [J]. Mater. Des., 2022, 213: 110326
39
Li Y, Martín D S, Wang J L, et al. A review of the thermal stability of metastable austenite in steels: Martensite formation [J]. J. Mater. Sci. Technol., 2021, 91: 200
doi: 10.1016/j.jmst.2021.03.020
40
Stormvinter A, Borgenstam A, Ågren J. Thermodynamically based prediction of the martensite start temperature for commercial steels [J]. Metall. Mater. Trans., 2012, 43A: 3870
41
Lee S J, Park K S. Prediction of martensite start temperature in alloy steels with different grain sizes [J]. Metall. Mater. Trans., 2013, 44A: 3423
42
Rahaman M, Mu W Z, Odqvist J, et al. Machine learning to predict the martensite start temperature in steels [J]. Metall. Mater. Trans., 2019, 50A: 2081
43
Wang C C, Zhu K Y, Hedström P, et al. A generic and extensible model for the martensite start temperature incorporating thermodynamic data mining and deep learning framework [J]. J. Mater. Sci. Technol., 2022, 128: 31
doi: 10.1016/j.jmst.2022.04.014
44
Ghosh G, Olson G B. Kinetics of F.C.C. → B.C.C. heterogeneous martensitic nucleation—I. The critical driving force for athermal nucleation [J]. Acta Metall. Mater., 1994, 42: 3361
45
Lu Q, Liu S L, Li W, et al. Combination of thermodynamic knowledge and multilayer feedforward neural networks for accurate prediction of MS temperature in steels [J]. Mater. Des., 2020, 192: 108696
46
Kannan R, Nandwana P. Accelerated alloy discovery using synthetic data generation and data mining [J]. Scr. Mater., 2023, 228: 115335
47
Zhang S J, Yi W, Zhong J, et al. Computer alloy design of Ti modified Al-Si-Mg-Sr casting alloys for achieving simultaneous enhancement in strength and ductility [J]. Materials, 2023, 16: 306
48
Zou H, Tian Y Y, Zhang L G, et al. Integrating machine learning and CALPHAD method for exploring low-modulus near-β-Ti alloys [J]. Rare Met., 2024, 43: 309
49
Fu H, Gao T C, Gao J B, et al. Breaking hardness and electrical conductivity trade-off in Cu-Ti alloys through machine learning and Pareto front [J]. Mater. Res. Lett., 2024, 12: 580
50
Zhang W L, Tang Y, Gao J H, et al. Determination of hardness and Young's modulus in fcc Cu-Ni-Sn-Al alloys via high-throughput experiments, CALPHAD approach and machine learning [J]. J. Mater. Res. Technol., 2024, 30: 5381
51
Liu X L, Zhang J X, Pei Z R. Machine learning for high-entropy alloys: Progress, challenges and opportunities [J]. Prog. Mater. Sci., 2023, 131: 101018
52
Xu B, Yin H Q, Jiang X, et al. Data-driven design of Ni-based turbine disc superalloys to improve yield strength [J]. J. Mater. Sci. Technol., 2023, 155: 175
doi: 10.1016/j.jmst.2023.01.032
53
Lu S, Zou M, Zhang X R, et al. Data-driven “cross-component” design and optimization of γ′-strengthened Co-based superalloys [J]. Adv. Eng. Mater., 2023, 25: 2201257
54
Trehern W, Ortiz-Ayala R, Atli K C, et al. Data-driven shape memory alloy discovery using Artificial Intelligence Materials Selection (AIMS) framework [J]. Acta Mater., 2022, 228: 117751
55
Zeng Y Z, Man M R, Bai K W, et al. Explore the full temperature-composition space of 20 quinary CCAs for FCC and BCC single-phases by an iterative machine learning + CALPHAD method [J]. Acta Mater., 2022, 231: 117865
56
Jin X Z, Luo H, Wang X F, et al. Data mining accelerated the design strategy of high-entropy alloys with the largest hardness based on genetic algorithm optimization [J]. MGE Adv., 2024, 2: e49
57
Vazquez G, Singh P, Sauceda D, et al. Efficient machine-learning model for fast assessment of elastic properties of high-entropy alloys [J]. Acta Mater., 2022, 232: 117924
58
Wen C, Wang C X, Zhang Y, et al. Modeling solid solution strengthening in high entropy alloys using machine learning [J]. Acta Mater., 2021, 212: 116917
59
Li Z, Nash W T, O'Brien S P, et al. cardiGAN: A generative adversarial network model for design and discovery of multi principal element alloys [J]. J. Mater. Sci. Technol., 2022, 125: 81
doi: 10.1016/j.jmst.2022.03.008
60
Zhang H T, Fu H D, Zhu S C, et al. Machine learning assisted composition effective design for precipitation strengthened copper alloys [J]. Acta Mater., 2021, 215: 117118
61
He J J, Li J J, Liu C B, et al. Machine learning identified materials descriptors for ferroelectricity [J]. Acta Mater., 2021, 209: 116815
62
Hartnett T Q, Sharma V, Garg S, et al. Accelerated design of MTX alloys with targeted magnetostructural properties through interpretable machine learning [J]. Acta Mater., 2022, 231: 117891
63
Wang C C, Zhang Z, Jing X Y, et al. Optimization of multistage femtosecond laser drilling process using machine learning coupled with molecular dynamics [J]. Opt. Laser Technol., 2022, 156: 108442
64
Zhang Z, Yang Z N, Wang C C, et al. Accelerating ultrashort pulse laser micromachining process comprehensive optimization using a machine learning cycle design strategy integrated with a physical model [J]. J. Intell. Manuf., 2024, 35: 449
65
Mondal B, Mukherjee T, DebRoy T. Crack free metal printing using physics informed machine learning [J]. Acta Mater., 2022, 226: 117612
66
Rao Z Y, Tung P Y, Xie R W, et al. Machine learning-enabled high-entropy alloy discovery [J]. Science, 2022, 378: 78
doi: 10.1126/science.abo4940
pmid: 36201584
67
Zou C X, Li J S, Wang W Y, et al. Integrating data mining and machine learning to discover high-strength ductile titanium alloys [J]. Acta Mater., 2021, 202: 211
68
Holm E A, Cohn R, Gao N, et al. Overview: Computer vision and machine learning for microstructural characterization and analysis [J]. Metall. Mater. Trans., 2020, 51A: 5985
69
Müller M, Stiefel M, Bachmann B I, et al. Overview: Machine learning for segmentation and classification of complex steel microstructures [J]. Metals, 2024, 14: 553
70
Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition [A]. 3rd International Conference on Learning Representations [C]. San Diego, May 7-9, 2015
71
Ronneberger O, Fischer P, Brox T. U-Net: Convolutional networks for biomedical image segmentation [A]. 18th International Conference on Medical Image Computing and Computer-Assisted Intervention [C]. Munich: Springer, 2015
72
He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition [A]. 2016 IEEE Conference on Computer Vision and Pattern Recognition [C]. Las Vegas: IEEE, 2016: 770
73
Shen C G, Wang C C, Huang M H, et al. A generic high-throughput microstructure classification and quantification method for regular SEM images of complex steel microstructures combining EBSD labeling and deep learning [J]. J. Mater. Sci. Technol., 2021, 93: 191
doi: 10.1016/j.jmst.2021.04.009
74
Kunselman C, Sheikh S, Mikkelsen M, et al. Microstructure classification in the unsupervised context [J]. Acta Mater., 2022, 223: 117434
75
Ma X D, Zhang Y Q, Wang C C, et al. Alloy microstructure segmentation through SAM and domain knowledge without extra training [J]. Scr. Mater., 2025, 260: 116581
76
Ren D, Wang C C, Wei X L, et al. Building a quantitative composition-microstructure-property relationship of dual-phase steels via multimodal data mining [J]. Acta Mater., 2023, 252: 118954
77
Zhao P L, Wang Y W, Jiang B Y, et al. Neural network modeling of titanium alloy composition-microstructure-property relationships based on multimodal data [J]. Mater. Sci. Eng., 2023, A879: 145202
78
Wang C C, Ren D, Li Y, et al. Prediction of deformation-induced martensite start temperature by convolutional neural network with dual mode features [J]. Materials, 2022, 15: 3495
79
Han S Y, Wang C C, Lai Q Q, et al. Fitting-free mechanical response prediction in dual-phase steels by crystal plasticity theory guided deep learning [J]. Acta Mater., 2025, Available online, 120936
80
Jin L C, Tan F X, Jiang S M. Generative adversarial network technologies and applications in computer vision [J]. Comput. Intel. Neurosci., 2020, 2020: 1459107
81
Goodfellow I J, Pouget-Abadie J, Mirza M, et al. Generative adversarial nets [A]. Proceedings of the 28th International Conference on Neural Information Processing Systems [C]. Montreal: MIT Press, 2014
82
Qian C, Tan R K, Ye W J. Design of architectured composite materials with an efficient, adaptive artificial neural network-based generative design method [J]. Acta Mater., 2022, 225: 117548
83
Narikawa R, Fukatsu Y, Wang Z L, et al. Generative adversarial networks-based synthetic microstructures for data-driven materials design [J]. Adv. Theory Simul., 2022, 5: 2100470
84
Long T, Zhang Y X, Fortunato N M, et al. Inverse design of crystal structures for multicomponent systems [J]. Acta Mater., 2022, 231: 117898
85
Yang X. A machine learning-based approach for materials microstructure analysis and prediction [D]. Houston City: Rice University, 2020
86
Cao Z H, Liu Q, Liu Q C, et al. A machine learning method to quantitatively predict alpha phase morphology in additively manufactured Ti-6Al-4V [J]. npj Comput. Mater., 2023, 9: 195
87
Yang Z J, Li X L, Brinson L C, et al. Microstructural materials design via deep adversarial learning methodology [J]. J. Mech. Des., 2018, 140: 111416
88
Kingma D P, Welling M. Auto-encoding variational Bayes [A]. 2nd International Conference on Learning Representations [C]. Banff, April 14-16, 2014
89
Pei Z R, Rozman K A, Doğan Ö N, et al. Machine-learning microstructure for inverse material design [J]. Adv. Sci., 2021, 8: 2101207
90
Ma X D, Zhang Y Q, Wang C C, et al. Creating a microstructure latent space with rich material information for multiphase alloy design [DB/OL]. arXiv: 2409. 02648, 2024
91
Jha D, Choudhary K, Tavazza F, et al. Enhancing materials property prediction by leveraging computational and experimental data using deep transfer learning [J]. Nat. Commun., 2019, 10: 5316
doi: 10.1038/s41467-019-13297-w
pmid: 31757948
92
Yamada H, Liu C, Wu S, et al. Predicting materials properties with little data using shotgun transfer learning [J]. ACS Cent. Sci., 2019, 5: 1717
93
Wei X L, van der Zwaag S, Jia Z X, et al. 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 [J]. Acta Mater., 2022, 235: 118103
94
Jiang L, Zhang Z H, Hu H, et al. A rapid and effective method for alloy materials design via sample data transfer machine learning [J]. npj Comput. Mater., 2023, 9: 26
95
Wei X L, Zhang C, Han S Y, et al. High cycle fatigue S-N curve prediction of steels based on transfer learning guided long short term memory network [J]. Int. J. Fatigue, 2022, 163: 107050
96
Cuomo S, Di Cola V S, Giampaolo F, et al. Scientific machine learning through physics-informed neural networks: Where we are and what's next [J]. J. Sci. Comput., 2022, 92: 88
97
Liu C, Wu H A. A variational formulation of physics-informed neural network for the applications of homogeneous and heterogeneous material properties identification [J]. Int. J. Appl. Mech., 2023, 15: 23500655
98
Jin H X, Zhang E R, Espinosa H D. Recent advances and applications of machine learning in experimental solid mechanics: A review [J]. Appl. Mech. Rev., 2023, 75: 061001
99
Zhang L J, Li K W, Wang H, et al. MFC-PINN: A method to improve the accuracy and robustness of acoustic emission source planar localization [J]. Measurement, 2024, 235: 114995
100
Liu T Y. AI for Science: Pursuing the brightest side of human intelligence [EB/OL]. (2023-01-05).
刘铁岩. AI for Science: 追求人类智能最光辉的一面[EB/OL]. (2023-01-05).
[1]
LI Dianzhong, HU Xiaoqiang, WANG Pei. Metal Chain Creation[J]. 金属学报, 2025, 61(2): 203-210.
