Machine Learning-Based Investigation of the Compressive Behavior of TI6AL4V Lattice Structures

Abstract:

This study aims to predict the effective Young’s modulus of Ti6Al4V lattice structures based on strut thickness and strut length using regression-based machine learning methods. Four predictive models, Linear Regression (LR), Polynomial Regression (PR), Support Vector Regression (SVR), and Gaussian Process Regression (GPR), were developed and compared in MATLAB using 25 sets of simulation data. Among these models, PR and GPR demonstrated the most promising performance in the actual versus predicted comparison, achieving the lowest root mean square error (RMSE) and the highest coefficient of determination (R²) for the given data size. Optimal training size for all models was around 36% to 52% of the total data set, which has a critical significance for data efficiency. To evaluate data sufficiency and model reliability, 5-fold Cross Validation was performed, and learning curves were generated to analyze how prediction error varies with the number of training samples. In the Learning curve, the PR model achieved its lowest RMSE, followed by GPR, which had the second best RMSE at its optimal training size. LR performed well for comparatively linear data, whereas SVR showed great variations and many shortcomings with the limited dataset.

• Periodical: Materials Science Forum (Volume 1198)
• Pages: 105-113
• DOI: https://doi.org/10.4028/p-7OSEze
• Online since: August 2026
• Authors: Maliha Binte Hasan*, Natasa Trisovic, Tamas Mankovits
• Keywords: Data Efficiency, Effective Young’s Modulus, Finite Element Analysis (FEA), Gaussian Process Regression, Machine Learning Regression, Polynomial Regression, Ti6Al4V Lattice Structures, Training Sample Optimization

References

[1] E. Marin et al., "Biomedical Applications of Titanium Alloys: A Comprehensive Review," Jan. 01, 2024, Multidisciplinary Digital Publishing Institute (MDPI).

DOI: 10.3390/ma17010114

[2] Y. W. Cui et al., "Towards load-bearing biomedical titanium-based alloys: From essential requirements to future developments," Nov. 2024, Elsevier Ltd.

DOI: 10.1016/j.pmatsci.2024.101277

[3] R. Alkentar et al., "Investigation of the Performance of Ti6Al4V Lattice Structures Designed for Biomedical Implants Using the Finite Element Method," Materials (Basel)., vol. 15, no. 18, Nov. 2022.

DOI: 10.3390/ma15186335

[4] L. Zhang et al., "A topology strategy to reduce stress shielding of additively manufactured porous metallic biomaterials," Int. J. Mech. Sci., vol. 197, Nov. 2021.

DOI: 10.1016/j.ijmecsci.2021.106331

[5] D. Savio et al., "When the Total Hip Replacement Fails: A Review on the Stress-Shielding Effect," Nov. 2022, MDPI.

DOI: 10.3390/pr10030612

[6] O. Bittredge et al., "Fabrication and Optimisation of Ti-6Al-4V Lattice-Structured Total Shoulder Implants Using Laser Additive Manufacturing," Materials (Basel)., vol. 15, no. 9, Nov. 2022.

DOI: 10.3390/ma15093095

[7] D. P. Papazoglou et al., "In Vitro Proliferation of MG-63 Cells in Additively Manufactured Ti-6Al-4V Biomimetic Lattice Structures with Varying Strut Geometry and Porosity," Materials (Basel)., vol. 17, no. 18, Nov. 2024.

DOI: 10.3390/ma17184608

[8] U. O. Agwu et al., "Assessing Tetrahedral Lattice Parameters for Engineering Applications through Finite Element Analysis," 3D Print. Addit. Manuf., vol. 8, no. 4, p.238–252, Nov. 2021.

DOI: 10.1089/3dp.2020.0222

[9] J. Yang et al., "Machine learning assisted prediction and analysis of in-plane elastic modulus of hybrid hierarchical square honeycombs," Thin-Walled Struct., vol. 198, Nov. 2024.

DOI: 10.1016/j.tws.2024.111736

[10] C. P. Karri et al., "Finite Element Analysis Approach for Optimal Design and Mechanical Performance Prediction of Additive Manufactured Sandwich Lattice Structures," J. Inst. Eng. Ser. D, 2024.

DOI: 10.1007/s40033-024-00650-7

[11] J. Bai, M. Li et al., "Prediction of Mechanical Properties of Lattice Structures: An Application of Artificial Neural Networks Algorithms," Materials (Basel)., vol. 17, no. 17, Nov. 2024.

DOI: 10.3390/ma17174222

[12] R. C. F. da Paixão et al., "Comparison of machine learning techniques to predict the compressive strength of concrete and considerations on model generalization," Rev. IBRACON Estruturas e Mater., vol. 15, no. 5, 2022.

DOI: 10.1590/S1983-41952022000500003

[13] N. Liu et al., "Estimation of static Young's modulus of sandstone types: effective machine learning and statistical models," Earth Sci. Informatics, vol. 17, no. 5, p.4339–4359, Nov. 2024.

DOI: 10.1007/s12145-024-01392-6

[14] A. V Tatachar, "Comparative Assessment of Regression Models Based On Model Evaluation Metrics," Int. Res. J. Eng. Technol., 2021, [Online]. Available: www.irjet.net.

[15] H. Liu et al., "Active learning for regression of structure-property mapping: the importance of sampling and representation," Digit. Discov., vol. 3, no. 10, p.1997–2009, Aug. 2024.

DOI: 10.1039/d4dd00073k

[16] A. Shmuel et al., "A comprehensive benchmark of machine and deep learning models on structured data for regression and classification," Neurocomputing, vol. 655, Nov. 2025.

DOI: 10.1016/j.neucom.2025.131337

[17] P. Xu et al., "Small data machine learning in materials science," Nov. 2023, Nature Research.

DOI: 10.1038/s41524-023-01000-z

[18] Y. Liu et al., "REVIEW MATERIALS SCIENCE Data quantity governance for machine learning in materials science".

[19] Q. Tao et al., "Machine learning strategies for small sample size in materials science," Nov. 2025, Science China Press.

DOI: 10.1007/s40843-024-3204-5

[20] S. Bishnoi et al., "Predicting Young's modulus of oxide glasses with sparse datasets using machine learning," J. Non. Cryst. Solids, vol. 524, Nov. 2019.

DOI: 10.1016/j.jnoncrysol.2019.119643

[21] D. L. Naik et al., "A novel sensitivity-based method for feature selection," J. Big Data, vol. 8, no. 1, Nov. 2021.

DOI: 10.1186/s40537-021-00515-w