Scholarlink

<p>Additive manufacturing advancements allow the manufacturing of mechanical metamaterials. However, modifications in parameters such as thickness, dimensions, or material composition enables the generation of numerous configurations, each influencing the mechanical properties. Therefore, accurately determining these mechanical properties is fundamental for novel materials and applications. Synergy between computational simulation and deep learning has enabled the prediction of mechanical properties. In this study, a 3D convolutional neural network is proposed to predict stress and strain in mechanical specimens. The dataset used to train and validate the model was generated through finite element analysis simulations, following tensile testing of printed cellular specimens under ASTM standard. These re-entrant honeycomb specimens were manufactured with varying wall thicknesses to compare FEA results with 3D CNN predictions. The results demonstrate that the 3D CNN model can accurately predict mechanical behavior, even under test conditions not included in the training or validation datasets.</p> Graphical abstract <p></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

Prediction of mechanical properties in 3D printed re-entrant honeycomb using 3D CNN

  • 期刊论文

  • 发表日期：

    2026年1月5日

查看全文
  • J. A. Hernández-Medina,
  • C. G. Flores-Hernández,
  • V. González-Huitrón,
  • G. A. Fonseca-Hernández,
  • G. Pineda-Chacón,
  • J. López-Barroso

摘要

Additive manufacturing advancements allow the manufacturing of mechanical metamaterials. However, modifications in parameters such as thickness, dimensions, or material composition enables the generation of numerous configurations, each influencing the mechanical properties. Therefore, accurately determining these mechanical properties is fundamental for novel materials and applications. Synergy between computational simulation and deep learning has enabled the prediction of mechanical properties. In this study, a 3D convolutional neural network is proposed to predict stress and strain in mechanical specimens. The dataset used to train and validate the model was generated through finite element analysis simulations, following tensile testing of printed cellular specimens under ASTM standard. These re-entrant honeycomb specimens were manufactured with varying wall thicknesses to compare FEA results with 3D CNN predictions. The results demonstrate that the 3D CNN model can accurately predict mechanical behavior, even under test conditions not included in the training or validation datasets.

Graphical abstract