Prediction of Fatigue Life of Natural Rubber by Different Machine Learning Models

XIAO Ling; WU You-yun

doi:10.14028/j.cnki.1003-3726.2025.25.279

您当前的位置：

首页 >

文章列表页 >

Prediction of Fatigue Life of Natural Rubber by Different Machine Learning Models

更新时间：2026-03-02

- Prediction of Fatigue Life of Natural Rubber by Different Machine Learning Models
- Polymer Bulletin Vol. 39, Issue 3, Pages: 447-456(2026)
- 作者机构：
  
  濮阳职业技术学院　濮阳　 457001
  河南大学濮阳工学院　濮阳　 457001
- 作者简介：
- 基金信息：
- DOI：10.14028/j.cnki.1003-3726.2025.25.279
  CLC：
- Received：24 October 2025，
  
  Accepted：25 November 2025，
  
  Online First：12 January 2026，
  
  Published：20 March 2026
- 稿件说明：
移动端阅览
肖玲, 伍又云. 基于不同机器学习模型预测天然橡胶疲劳寿命. 高分子通报, 2026, 39(3), 447-456.

Xiao, L.; Wu, Y. Y. Prediction of fatigue life of natural rubber by different machine learning models. Polym. Bull. (in Chinese), 2026, 39(3), 447-456.
肖玲, 伍又云. 基于不同机器学习模型预测天然橡胶疲劳寿命. 高分子通报, 2026, 39(3), 447-456. DOI： 10.14028/j.cnki.1003-3726.2025.25.279.

Xiao, L.; Wu, Y. Y. Prediction of fatigue life of natural rubber by different machine learning models. Polym. Bull. (in Chinese), 2026, 39(3), 447-456. DOI： 10.14028/j.cnki.1003-3726.2025.25.279.

摘要

用人工神经元网络(ANN)和径向基网络(RBF)对天然橡胶试样的疲劳寿命进行了建模和预测，利用敏感性分析定量研究了环境温度、橡胶硬度、峰值应变对橡胶疲劳寿命的影响程度。当环境温度为25、50和70 ℃，橡胶硬度为N40和N50，峰值应变在0.4~1.3之间时，试样疲劳寿命在9984~604247次之间。通过优化隐含层神经元数量，发现3-25-10-1结构的ANN网络和3-47-1结构的RBF网络预测疲劳寿命与试验结果均方误差(MSE)最低，分别为0.0383和0.0045，RBF网络预测准确性更高，性能曲线相关系数

均在0.9700以上，最高达到了0.9847，高于ANN的0.9511。峰值应变对疲劳寿命的影响最显著，达到了62.14%，高于环境温度的31.57%，橡胶硬度的影响最小，为6.29%。

Abstract

Artificial neural networks (ANN) and radial basis function networks (RBF) were employed to model and predict the fatigue life of natural rubber samples. Sensitivity analysis was conducted to quantitatively evaluate the influence of environmental temperature

rubber hardness

and peak strain on fatigue life. Under the experimental conditions (environmental temperature: 25

70 ℃; rubber hardness: N40

N50; peak strain: 0.4−1.3)

the fatigue life of the samples varied from 9984 to 604247 cycles. By optimizing the number of neurons in t

he hidden layers

we found that the ANN model with a 3-25-10-1 structure and the RBF model with a 3-47-1 structure yielded the lowest mean squared error (MSE) in fatigue life prediction

with values of 0.0383 and 0.0045

respectively. The RBF model exhibited higher prediction accuracy

with the correlation coefficient (

) between predicted and measured fatigue life exceeding 0.9700 and reaching a maximum of 0.9847—superior to the ANN model (

=0.9511). Among the three factors

peak strain exerted the most significant influence on fatigue life (contribution: 62.14%)

followed by environmental temperature (31.57%)

whereas rubber hardness had the least impact (6.29%).

关键词

Keywords

references

Hu, W. Y. ; Lu, W. T. ; Fei, F. ; Dai, W. S. ; Chai, X. ; Zhou, P. L. ; Wang, J. C . Self-healing flexible sensor based on epoxidized natural rubber with the synergistic effect of coordination and hydrogen bonds . Langmuir , 2024 , 40 ( 32 ), 17009 – 17019 .

Yao, J. J. ; Zheng, Y. F. ; Tang, J. Z. ; Wang, S. Q. ; Luo, Y. Z. ; Yang, C . Multi-layered solid element for seismic analysis of rubber bearings considering stress continuity based on finite particle method . Eng. Struct. , 2024 , 318 , 118713 .

Payungwong, N. ; Wu, J. R. ; Sakdapipanich, J . Unlocking the potential of natural rubber: a review of rubber particle sizes and their impact on properties . Polymer , 2024 , 308 , 127419 .

Belkhiria, S. ; Abroug, A. ; Fathallah, R . A simple fatigue methodology for filled natural rubbers under positive and negative R ratios . Mater. Res. Express , 2024 , 11 ( 5 ), 055307 .

Mouslih, Y. ; Le Cam, J. B. ; Ruellan, B. ; Jeanneau, I. ; Canévet, F . Fatigue life reinforcement of carbon black filled natural rubber under non-relaxing torsion loadings and comparison with non-relaxing tension loadings . Fatigue Fract. Eng. Mater. Struct. , 2024 , 47 ( 5 ), 1620 – 1637 .

Bakhshizade, A. ; Ghasemi-Ghalebahman, A. ; Ali Hajimousa, M . Development of novel models for fatigue life assessment of natural rubber/styrene-butadiene rubber/nanoclay nanocomposite . Polym. Compos. , 2024 , 45 ( 2 ), 1851 – 1871 .

刘湘楠 . 橡胶隔振器概率疲劳寿命评估方法研究 . 华南理工大学博士学位论文 , 2023 .

赖威 . 橡胶隔振器疲劳寿命预测与试验方法研究 . 华南理工大学硕士学位论文 , 2022 .

Liang, J. M. ; Zhang, H. ; Luo, Z. ; Zhong, J. C. ; Wang, X. Y . Fatigue-crack-propagation behavior and mechanism of long hemp-fiber-enhanced styrene-butadiene rubber/carbon black compounds . Polym. Compos. , 2024 , 45 ( 4 ), 2941 – 2953 .

Luo, K. R . Fatigue prediction criteria for rubber antivibration design with temperature effects . Int. J. Fatigue , 2024 , 183 , 108271 .

Fang, Y. ; Sun, C. H. ; Zhu, Z. K. ; Zhang, G. F. ; Yang, H. ; Gao, W. W. ; Liu, X. T . Failure analysis for air spring systems of urban rail vehicles considering load spectrum . Eng. Fail. Anal. , 2024 , 159 , 107997 .

刘超 , 罗传富 . 基于大变形疲劳理论的橡胶材料及制品疲劳寿命预测方法 . 高分子材料科学与工程 , 2024 , 40 ( 7 ), 103 – 111 .

Wang, A. H. ; Wang, H. ; Ling, C. W. ; Wang, T. ; Li, F. ; Yang, S. S. ; Zhao, M . Fatigue life prediction of rubber suspension bushings based on virtual road load spectrum pulses . Results Eng. , 2024 , 23 , 102693 .

沈兴铿 , 徐鹤鸣 , 刘伟 , 张屹尚 , 戴瑛 , 陈新民 , 周洪民 . P型橡胶卡箍的拉压疲劳性能测试和寿命预测 . 南京航空航天大学学报 , 2024 , 56 ( 3 ), 494 – 503 .

Wang, X. L. ; Liu, Z. Y. ; Zeng, X. K. ; Xu, Y . A method to obtain probabilistic fatigue lives under varying strain peaks for rubber components with small samples . Fatigue Fract. Eng. Mater. Struct. , 2024 , 47 ( 8 ), 2955 – 2968 .

Sheng, X. W. ; Liao, J. C. ; Zheng, W. Q. ; Zhu, Z. H. ; Lin, C . Fatigue characteristics of rubber isolation layers used in high speed railway ballastless tracks: a theoretical and experimental study . Int. J. Fatigue , 2024 , 186 , 108381 .

上官文斌 , 李明敏 , 段小成 . 动力总成橡胶悬置高温疲劳特性的预测与试验研究 . 振动与冲击 , 2015 , 34 ( 15 ), 66 – 71 .

Liu, X. N. ; Zhao, X. Z. ; Shangguan, W. B . Fatigue life prediction of natural rubber components using an artificial neural network . Fatigue Fract. Eng. Mater. Struct. , 2022 , 45 ( 6 ), 1678 – 1689 .

Choi, B. ; Lee, J. H. ; Kim, D. H . Solving local minima problem with large number of hidden nodes on two-layered feed-forward artificial neural networks . Neurocomputing , 2008 , 71 ( 16-18 ), 3640 – 3643 .

Nasrollahnejad, A. ; Rajabi, H. ; Nasrollahnejad, M. B. ; Tahmasebi, S . Simulation of peak ground acceleration by artificial neural network and radial basis function network . Int. J. Mod. Educ. Comput. Sci. , 2014 , 6 ( 10 ), 10 - 15 .

Nazir, K. ; Memon, S. A. ; Saurbayeva, A . A novel framework for developing a machine learning-based forecasting model using multi-stage sensitivity analysis to predict the energy consumption of PCM-integrated building . Appl. Energy , 2024 , 376 , 124180 .

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Natural Rubber Crosslinking Structure and Homogenization Stress Affected by Zinc 1,5-Pentylidene Dithiocarbamate

Influence of Diatomite on the Properties of Silica/NR/SSBR Composites

Effect of None Rubber Components on Relationship between Structure and Performance of Natural Rubber

Study on Preparation of Hemp Fiber Reinforced Natural Rubber and Thermogenesis

Related Author

WANG Zhi-qiang

XUE Ao-nan

ZHU Bo-cong

LIU Da-chen

XIAO Xiao-kun

Pei-pei LI

Zhao-hong ZHANG

Li-hua XING

Related Institution

School of Materials Science and Engineering, Shenyang University of Chemical Technology

Weilin New Materials Technology Co., Ltd.

West Anhui University

Xuzhou College of Industrial Technology

Analytical Instrumentation Center, College of Chemistry and Molecular Engineering, Peking University

⁰