Liu, J;
Chang, J;
Yu, J;
Zhang, W;
Huang, S;
(2023)
Machine learning-based optimization design of bistable curved shell structures with variable thickness.
Structures
, 52
pp. 175-186.
10.1016/j.istruc.2023.03.124.
Preview |
Text
Manuscript.pdf - Accepted Version Download (1MB) | Preview |
Abstract
The mechanical performance of curved shell structures is difficult to predict due to their complex geometric nonlinearity. There have been many efforts to improve the mechanical property of curved shell structures by designing the thickness distributions. However, due to the nonlinear characteristics of variable-thickness shells, it is impossible to exhaust all possible structural forms through experimental and theoretical approaches. In this paper, we report a new machine learning (ML)-based approach to design and optimize bistable curved shell structures. The ML model is used to establish the underlying mapping relationship between structural parameters and specific performance. Nonlinear shell structures are efficiently and accurately designed and optimized with optimal backward snapping forces (B). The results demonstrate an effective approach for the design and optimization of curved shell structures and provide a valuable reference for future study of nonlinear structures. Moreover, this approach presents an efficient means of designing advanced metamaterials. The modular design of meta-atoms based on ML has the potential to construct metamaterials with specific properties and functionalities, which will lead to extensive applications in different fields.
Archive Staff Only
 |
View Item |
