| Autor: |
Canhui Lin, Ke Xu, Yingguang Li, Xu Liu, Chenli Zhou |
| Jazyk: |
angličtina |
| Rok vydání: |
2025 |
| Předmět: |
|
| Zdroj: |
Materials & Design, Vol 249, Iss , Pp 113529- (2025) |
| Druh dokumentu: |
article |
| ISSN: |
0264-1275 |
| DOI: |
10.1016/j.matdes.2024.113529 |
| Popis: |
Lightweight structures are ubiquitous in nature and extensively applied in high-end industries for load-bearing purpose. Parametric design is superior for its rapid generation of complex geometries controlled by user-specified parameters, thus has been increasingly explored in structural optimization. Conventional parametric design approaches for lightweight structure are primarily focus on size and shape, while the topology still requires lengthy iterative process. The major issues that hinder parametric control of topological design include singularity of the stress field and the lack of proper operators for discretized geometry. This paper introduced an implicit neural representation for parametric design of stress-oriented cellular structure, which exhibited great potential to resolve the aforementioned issues by harnessing the universal approximation and resolution invariant capability of a neural network. The structure was implicitly trained to smoothly align with the principal stress field of the input geometry under arbitrary loading condition. A tailored wave projection function together with morphological operators were employed for parametric control of the output cellular structure. The proposed framework enables the parametric design and fabrication of various cellular structures, resulting in enhanced load-bearing capacity compared to structures designed using conventional methods. This advancement opens up new possibilities for the effective design of lightweight structures. |
| Databáze: |
Directory of Open Access Journals |
| Externí odkaz: |
|
Full Text from ScienceDirect