Wang, Hao, Ren, Yuanfei, Kadir, Mohammed Rafiq Abdul, Zhu, Hanxing  ORCID: https://orcid.org/0000-0002-3209-6831 and Lyu, Yongtao
2027.
Data-driven multi-objective inverse design of programmable Vorochiral metamaterials for customized orthotic midsoles.
Thin-Walled Structures
232
(Part 1)
, 115667.
10.1016/j.tws.2026.115667
|
Preview |
PDF
- Accepted Post-Print Version
Available under License Creative Commons Attribution. Download (1MB) | Preview |
Abstract
The customized design of advanced load-bearing components, such as foot orthoses, remains a critical engineering challenge due to the highly nonlinear relationships between architected metamaterial configurations and macroscopic mechanical performance. In this study, a data-driven multi-objective inverse design framework was proposed to optimize programmable Vorochiral mechanical metamaterials for customized orthotic midsoles. Guided by plantar pressure distributions, an infilling design strategy for plantar biomechanical modulation was implemented across the forefoot, midfoot, and heel regions to precisely mitigate high-pressure concentrations. Optimal Latin hypercube sampling and a variational autoencoder were employed to construct and augment the dataset, enabling the training of a robust backpropagation neural network (BPNN) surrogate model for efficient performance prediction under limited sample conditions. By integrating the BPNN with the non-dominated sorting genetic algorithm II, the Pareto front was efficiently extracted, and the optimal configuration parameters were identified using a comprehensive decision-making framework based on the entropy weight-TOPSIS method. The designed Vorochiral orthotic midsole was systematically evaluated through finite element simulations, mechanical experiments, as well as static standing and dynamic gait tests. The results demonstrated that, compared with midsoles infilled with conventional structures, the optimized Vorochiral midsole significantly reduced both peak and mean plantar pressures, achieving reductions of up to 33.24% and 37.94%, respectively. These findings highlight its strong potential for customized biomechanical interventions. Furthermore, the proposed framework establishes a generalizable data-driven paradigm for inverse design and performance modulation of architected metamaterials, facilitating efficient customized engineering design across broader application domains.
| Item Type: | Article |
|---|---|
| Date Type: | Publication |
| Status: | Published |
| Schools: | Schools > Engineering |
| Additional Information: | RRS policy applied |
| Publisher: | Elsevier |
| ISSN: | 0263-8231 |
| Date of First Compliant Deposit: | 23 September 2026 |
| Date of Acceptance: | 11 September 2026 |
| Last Modified: | 23 Sep 2026 11:00 |
| URI: | https://orca.cardiff.ac.uk/id/eprint/189755 |
Actions (repository staff only)
 |
Edit Item |
Altmetric
Altmetric