Application of transfer learning in the prediction of elastoplastic response of short fiber reinforced polymers

doi:10.19936/j.cnki.2096-8000.20260228.002

Abstract

Abstract: With the wide application of composite materials in aerospace and automotive fields, it has become more and more important to accurately predict their mechanical properties. In this study, a transfer learning-based stress-strain curve prediction method for short fiber reinforced polymers was proposed. Firstly, the database was constructed by DIGIMAT and the Latin hypercube sampling technique was used to select samples to improve the efficiency of model training. Then, using artificial neural network (ANN) as a surrogate model, and through the transfer learning method, the stress-strain prediction model of the new material can be quickly obtained. The results show that the transfer learning model can effectively capture the key features of the stress-strain behavior of materials, especially in predicting the fifth-order polynomial coefficients. The effects of fiber volume fraction and aspect ratio on the mechanical properties of materials were further analyzed, and it was found that the fibers with larger length-diameter ratios could transmit stress more effectively. This study provides an effective tool for the performance prediction of short fiber reinforced polymers, especially when data acquisition is difficult or the cost is high, the potential and application prospects of transfer learning in the performance prediction of composite materials are demonstrated.

Key words: short fiber reinforced polymers, transfer learning, stress-strain curves, artificial neural networks

CLC Number:

TB332

XING Wenqi, ZHU Shuiwen, WU Shunxin. Application of transfer learning in the prediction of elastoplastic response of short fiber reinforced polymers[J]. COMPOSITES SCIENCE AND ENGINEERING, 2026, 0(2): 10-19.

References

[1] 蔡四维. 短纤维复合材料理论与应用[M]. 北京: 人民交通出版社, 1994.
[2] MOKARIZADEHHAGHIGHISHIRAZI M, BUFFEL B, LOMOV S V, et al. Homogenisation of the local thermal conductivity in injection-moulded short fibre reinforced composites[J]. Polymers, 2022, 14(16): 3360.
[3] CHOU T W, NOMURA S, TAYA M. A self-consistent approach to the elastic stiffness of short-fiber composites[J]. Journal of Composite Materials, 1980, 14(3): 178-188.
[4] MORI T, TANAKA K. Average stress in matrix and average elastic energy of materials with misfitting inclusions[J]. Acta Metallurgica, 1973, 21(5): 571-574.
[5] HUANG Z M, ZHANG C C, XUE Y D. Stiffness prediction of short fiber reinforced composites[J].International Journal of Mechanical Sciences, 2019, 161/162: 105068.
[6] HUANG Z M. Micromechanical prediction of ultimate strength of transversely isotropic fibrous composites[J]. International Journal of Solids & Structures, 2001, 38(22/23): 4147-4172.
[7] HAMID Z M A, FINDEISEN C, LAVEUVE D M, et al. A continuum damage mechanics model for fatigue degradation and failure of short fiber reinforced composites[J]. Composites Science and Technology, 2024, 254: 110679.
[8] NCIRI M, NOTTA-CUVIER D, LAURO F, et al. Modelling and characterisation of dynamic behaviour of short-fibre-reinforced composites[J]. Composite Structures, 2017, 160: 516-528.
[9] MIRKHALAF S, EGGELS E, BEURDEN T V, et al. A finite element based orientation averaging method for predicting elastic properties of short fiber reinforced composites[J]. Composites Part B: Engineering, 2020, 202: 108388.
[10] HUANG H B, HUANG Z M. Tensile failure prediction of a short fiber or particle composite[J]. Polymer Composites, 2021, 42(2): 774-790.
[11] MENTGES N, DASHTBBOZORG B, MIRKHALAF M. A micromechanics-based artificial neural networks model for elastic properties of short fiber composites[J]. Composites Part B: Engineering, 2021, 213: 108736.
[12] 王雅哲, 马其华. 机器学习在复合材料领域中的应用进展[J]. 工程塑料应用, 2023, 51(9): 167-174.
[13] LIU X, TIAN S, TAO F, et al. A review of artificial neural networks in the constitutive modeling of composite materials[J]. Composites Part B: Engineering, 2021, 224: 109152.
[14] YANG C, KIM Y, RYU S, et al. Prediction of composite microstructure stress-strain curves using convolutional neural networks[J]. Materials and Design, 2025, 189: 108509.
[15] QIU C, HAN Y Z, SHANMUGAM L, et al. A deep learning-based composite design strategy for efficient selection of material and layup sequences from a given database[J]. Composites Science and Technology, 2021, 230: 109154.
[16] 段尊义, 刘亿, 张皓翔, 等. 基于机器学习的变刚度纤维增强复合材料最小化结构柔顺性优化设计[J]. 力学学报, 2024, 56(7): 1849-1860.
[17] 王吉玲, 金浩, 郭瑞文, 等. 基于机器学习的短纤维增强复合材料弹性力学性能预测[J]. 复合材料学报, 2024, 41(11): 6261-6272.
[18] TAO F, LIU X, DU H D, et al. Discover failure criteria of composites from experimental data by sparse regression[J]. Composites Part B: Engineering, 2022, 239: 109947.
[19] BHUIYAN F H, KOTTHOFF L, FERTIG R. A machine learning technique to predict biaxial failure envelope of unidirectional composite lamina[J]. American Society for Composites, 2018(3): 1451-1463.
[20] MUKHERJEE A, SCHMAUDER S, RUHLE M. Artificial neural networks for the prediction of mechanical behavior of metal matrix composites[J]. Acta Metallurgica et Materialia, 1995, 43(11): 4083-4091.
[21] LEE C S, HWANG W, PARK H C, et al. Failure of carbon/epoxy composite tubes under combined axial and torsional loading 1. Experimental results and prediction of biaxial strength by the use of neural networks[J]. Composites Science and Technology, 1999, 59(12): 1779-1788.
[22] KIM Y, KIM Y, YANG C, et al. Deep learning framework for material design space exploration using active transfer learning and data augmentation[J]. npj Computational Materials, 2021, 7(1): 140.
[23] 冯彦皓, 俞自涛, 陆江, 等. 基于模型迁移学习的木材导热系数预测[J]. 工程热物理学报, 2024, 45(3): 865-872.
[24] ZHANG Z Y, LIU Q Y, WU D Z. Predicting stress-strain curves using transfer learning: knowledge transfer across polymer composites[J]. Materials & Design, 2022, 218: 110700.
[25] 沈观林, 胡更开, 刘彬. 复合材料力学: 第2版[M]. 北京: 清华大学出版社, 2013.
[26] MAHARANA S K, SONI G, MITRA M. A machine learning based prediction of elasto-plastic response of a short fiber reinforced polymer (SFRP) composite[J]. Modelling and Simulation in Materials Science and Engineering, 2023, 31(7): 075001.
[27] KETKAR N, MOOLAYIL J. Deep learning with Python: learn best practices of deep learning models with PyTorch[M]. Berkeley: Apress Berkeley, 2021: 27-91.