Generative machine learning framework for inverse design of matamaterials based on prescribed mechanical behavior

Chan Soo Ha, Desheng Yao, Chenang Liu, Zhenyu Kong, Han Liu, Mathieu Bauchy, Xiaoyu ZhengZenodo 2023 · DOI

Abstract

This repository contains a source code and dataset used to generate results that are reported in the paper entitled of "Rapid Inverse Design of Metamaterials based on Prescribed Mechanical Behavior through Machine Learning". A generative machine learning framework utilizes an inverse prediction and forward validation modules where each module is composed of five distinct neural network models. The framework offers a rapid inverse design of metamaterials in response to user-defined uniaxial compressive mechanical behavior. The framework can be readily extended to inverse-design other mechanical behaviors (e.g., tension, shear, torsion, bending, etc.) when the input follows a similar form of the stress-strain curve used in the paper.
BibTeX
@inproceedings{ha2023generative,
  author = {Chan Soo Ha and Desheng Yao and Chenang Liu and Zhenyu Kong and Han Liu and Mathieu Bauchy and Xiaoyu Zheng},
  title = {{Generative machine learning framework for inverse design of matamaterials based on prescribed mechanical behavior}},
  booktitle = {Zenodo},
  year = {2023},
  doi = {10.5281/zenodo.8210037},
}

Loading related content...

Figure