A new framework for predicting tensile stress of natural rubber based on data augmentation and molecular dynamics simulation data

Abstract

This study addresses the challenge of predicting the tensile stress of natural rubber with limited molecular dynamics simulation data, which is a crucial mechanical property for this material. Molecular dynamics (MD) simulations are limited by their scale and computational cost, making it difficult to obtain sufficient data to train machine learning algorithms. To overcome this limitation, we propose a machine learning framework involving three stages: (1) utilizing a Variational Autoencoder (VAE) to rapidly expand the data diversity; (2) employing Ordinary Kriging (OK) to label the VAE-generated virtual samples; and (3) training gradient enhanced regression [Gradient Boosting Regression (GBR)] models by using relevant data on tensile stress in natural rubber. The results demonstrate that the generated data exhibits enhanced rationality, significantly improving the accuracy and reliability of various regression models. This approach provides an effective solution to the problem of data scarcity in MD simulations.

Keywords

Natural rubber / tensile stress / Variational Autoencoder / Ordinary Kriging / data augmentation algorithms

References

[1] Chibani S.Machine learning approaches for the prediction of materials properties.APL Mater2020;8:080701

[2] Li D,Chen Z,Dong Y.Accelerating the design and development of polymeric materials via deep learning: current status and future challenges.APL Mach Learn2023;1:021501

[3] Geng X,Wu HH.Data-driven and artificial intelligence accelerated steel material research and intelligent manufacturing technology.Mater Genome Eng Adv2023;1:e10

[4] Shi R,Song X.Tensile performance and viscoelastic properties of rubber nanocomposites filled with silica nanoparticles: a molecular dynamics simulation study.Chem Eng Sci2023;267:118318

[5] Yagyu H.Coarse-grained molecular dynamics simulation of the effects of strain rate on tensile stress of cross-linked rubber.Soft Mater2015;13:263-70

[6] Frankland SJV, Harik VM, Odegard GM, Brenner DW, Gates TS. The stress-strain behavior of polymer-nanotube composites from molecular dynamics simulation.Compos Sci Technol2003;63:1655-61

[7] Odegard GM,Gowtham S,He J.Predicting mechanical response of crosslinked epoxy using ReaxFF.Chem Phys Lett2014;591:175-8

[8] Chung I,Cho M.A neural network constitutive model for hyperelasticity based on molecular dynamics simulations.Numer Meth Eng2021;122:5-24

[9] Lee J,Kim JH.Importance of cohesive failure mode in fracture toughness enhancement of polymer nanocomposites with covalent grafting: a multiscale study.Compos Part B Eng2024;279:111462

[10] Wang H,Kim JH.Multiscale strategy to predict the fracture toughness and crack extension behavior of ozone-functionalized carbon nanotube/epoxy nanocomposites.Chem Eng J2023;465:142985

[11] Chen Q,Duan P.Manipulating the mechanical properties of cis-polyisoprene nanocomposites via molecular dynamics simulation.Polymer2022;256:125233

[12] Chen Q,Huang Y.Structure-mechanics relation of natural rubber: insights from molecular dynamics simulations.ACS Appl Polym Mater2022;4:3575-86

[13] Le Cam JB, Huneau B, Verron E, Gornet L. Mechanism of fatigue crack growth in carbon black filled natural rubber.Macromolecules2004;37:5011-7

[14] Trabelsi S,Rault J.Stress-induced crystallization around a crack tip in natural rubber.Macromolecules2002;35:10054-61

[15] Shen L.A virtual sample generation algorithm supporting machine learning with a small-sample dataset: a case study for rubber materials.Comput Mater Sci2022;211:111475

[16] Lockner Y,Zhao W.Transfer learning with artificial neural networks between injection molding processes and different polymer materials.J Manuf Process2022;73:395-408

[17] Chen Z,Liu J.Application of Gaussian processes and transfer learning to prediction and analysis of polymer properties.Comput Mater Sci2023;216:111859

[18] Kim Y,Yang C,Gu GX.Deep learning framework for material design space exploration using active transfer learning and data augmentation.npj Comput Mater2021;7:609

[19] Akdemir A.Research on task discovery for transfer learning in deep neural networks. In: Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: Student Research Workshop. Association for Computational Linguistics; 2020. pp. 33-41.

[20] Wu X,Aickelin U.A Bayesian approach to (online) transfer learning: theory and algorithms.Artif Intell2023;324:103991

[21] Zhang Y,Li L.Thermal stability prediction of copolymerized polyimides via an interpretable transfer learning model.J Mater Inf2024;4:8

[22] Xu P,Li M.Virtual sample generation in machine learning assisted materials design and discovery.J Mater Inf2023;3:16

[23] Li D,Liu J.NNI-SMOTE-XGBoost: A novel small sample analysis method for properties prediction of polymer materials.Macromol Theory Simul2021;30:2100010

[24] Hu A,Chen Q.Glass transition of amorphous polymeric materials informed by machine learning.APL Mach Learn2023;1:026111

[25] Huang Y,Zhang Z.A machine learning framework to predict the tensile stress of natural rubber: based on molecular dynamics simulation data.Polymers2022;14:1897 PMCID:PMC9103449

[26] Radford A,Chintala S. Unsupervised representation learning with deep convolutional generative adversarial networks. arXiv. [Preprint.] Jan 7, 2016 [accessed 2024 Aug 14]. Available from: https://arxiv.org/abs/1511.06434.

[27] Nguyen A,Yosinski J,Clune J. Synthesizing the preferred inputs for neurons in neural networks via deep generator networks. In: NIPS’16: Proceedings of the 30th International Conference on Neural Information Processing Systems. Curran Associates Inc.; 2016. pp. 3395-403. Available from: https://dl.acm.org/doi/abs/10.5555/3157382.3157477. [Last accessed on 14 Aug 2024]

[28] Zhu JY,Shechtman E. Generative visual manipulation on the natural image manifold. arXiv. [Preprint.] Dec 16, 2018 [accessed 2024 Aug 14]. Available from: https://arxiv.org/abs/1609.03552.

[29] Denton E,Szlam A. Deep generative image models using a Laplacian pyramid of adversarial networks. In: Proceedings of the 28th International Conference on Neural Information Processing Systems - Volume 1. Cambridge, MA, USA: MIT Press; 2015. pp. 1486-94. Available from: https://dl.acm.org/doi/10.5555/2969239.2969405. [Last accessed on 14 Aug 2024]

[30] Chen Q,Huang Y.Predicting natural rubber crystallinity by a novel machine learning algorithm based on molecular dynamics simulation data.Langmuir2023;39:17088-99

[31] Arjovsky M. Towards principled methods for training generative adversarial networks. arXiv. [Preprint.] Jan 17, 2017 [accessed 2024 Aug 14]. Available from: https://arxiv.org/abs/1701.04862.

[32] Kingma DP. Auto-encoding variational bayes. arXiv. [Preprint.] Dec 10, 2022 [accessed 2024 Aug 14]. Available from: https://arxiv.org/abs/1312.6114.

[33] Kulkarni TD,Kohli P. Deep convolutional inverse graphics network. arXiv. [Preprint.] Jun 22, 2015 [accessed 2024 Aug 17]. Available from: https://arxiv.org/abs/1503.03167.

[34] Gregor K,Graves A,Wierstra D. DRAW: a recurrent neural network for image generation. arXiv. [Preprint.] May 20, 2015 [accessed 2024 Aug 14]. Available from: https://arxiv.org/abs/1502.04623.

[35] Sønderby CK,Maaløe L,Winther O. Ladder variational autoencoders. In: NIPS’16: Proceedings of the 30th International Conference on Neural Information Processing Systems. Red Hook, NY, USA: Curran Associates Inc.; 2016. pp. 3745-53. Available from: https://dl.acm.org/doi/abs/10.5555/3157382.3157516. [Last accessed on 14 Aug 2024]

[36] Im DI,Memisevic R.Denoising criterion for variational auto-encoding framework.AAAI2017;31:1

[37] Wang X.Data supplement for a soft sensor using a new generative model based on a variational autoencoder and Wasserstein GAN.J Process Control2020;85:91-9

[38] Ohno H.Auto-encoder-based generative models for data augmentation on regression problems.Soft Comput2020;24:7999-8009

[39] Kulosa M,Boeff M,Schmidt V.A study on microstructural parameters for the characterization of granular porous ceramics using a combination of stochastic and mechanical modeling.Int J Appl Mechan2017;09:1750069

[40] Díaz-montiel P,Venkataraman S.Kriging and dimension reduction techniques for delamination detection in composites using electrical resistance tomography.Eng Optim2023;55:19-34

[41] Kookalani S,Torres JLC.Structural performance assessment of GFRP elastic gridshells by machine learning interpretability methods.Front Struct Civ Eng2022;16:1249-66

[42] Daya AA.A comparative study between simple kriging and ordinary kriging for estimating and modeling the Cu concentration in Chehlkureh deposit, SE Iran.Arab J Geosci2015;8:6003-20

[43] Massey FJ Jr.The Kolmogorov-Smirnov test for goodness of fit.J Am Stat Assoc1951;46:68-78

[44] Lundberg S. A unified approach to interpreting model predictions. arXiv. [Preprint.] Nov 25, 2017 [accessed 2024 Aug 14]. Available from: https://arxiv.org/abs/1705.07874.