Scholarlink

<p>To improve the adaptability of creep constitutive models to advanced materials and extreme environments, this paper proposes a general artificial neural network (ANN) creep model that can predict time-dependent creep behaviors. By leveraging datasets from the stress-based and strain-based models, the ANN model is embedded into the Abaqus creep subroutine with the predicted value and chain differentiation rule. To further illustrate the current model, the Liu–M model and the modified Wen–Tu model are employed for comparison. The results highlight the ANN model’s effectiveness in simulating creep deformation and crack propagation while emphasizing the influence of dataset quality, mesh size, and incremental step size on predictive accuracy. Von Mises stress predicted by the ANN model decreases with the increasing damage parameters, which is consistent with the trends of the modified Wen–Tu model. Furthermore, creep crack growth (CCG) analysis using the ANN model shows strong consistency with the published literature, where the predicted C* parameters under 1500 N and 2000 N loads almost entirely fall within the 95% prediction bands. This paper presents a flexible and robust methodology for conducting creep analysis, providing a foundation for further research and practical implementation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning-based constitutive modeling for creep behaviors: incorporating damage evolution parameter

  • 期刊论文

  • 发表日期：

    2026年3月5日

查看全文
  • Zheng Man,
  • Baoyin Zhu,
  • Zhigang Liang,
  • Jun Hong,
  • Shuitao Gu,
  • Gongye Zhang

摘要

To improve the adaptability of creep constitutive models to advanced materials and extreme environments, this paper proposes a general artificial neural network (ANN) creep model that can predict time-dependent creep behaviors. By leveraging datasets from the stress-based and strain-based models, the ANN model is embedded into the Abaqus creep subroutine with the predicted value and chain differentiation rule. To further illustrate the current model, the Liu–M model and the modified Wen–Tu model are employed for comparison. The results highlight the ANN model’s effectiveness in simulating creep deformation and crack propagation while emphasizing the influence of dataset quality, mesh size, and incremental step size on predictive accuracy. Von Mises stress predicted by the ANN model decreases with the increasing damage parameters, which is consistent with the trends of the modified Wen–Tu model. Furthermore, creep crack growth (CCG) analysis using the ANN model shows strong consistency with the published literature, where the predicted C* parameters under 1500 N and 2000 N loads almost entirely fall within the 95% prediction bands. This paper presents a flexible and robust methodology for conducting creep analysis, providing a foundation for further research and practical implementation.