We are pleased to share our most recent publication, “Inverse Multi-Objective Design of Three-Dimensional Plate-Based Heterogeneous Mechanical Metamaterials” in the International Journal of Mechanical Sciences.
In this research, we developed a data-efficient, clustering-aware design framework that combines deep neural networks with genetic algorithms and finite element simulations to generate architected structures with tailored mechanical properties automatically. Our method enables simultaneous control of stiffness, energy absorption, and relative density, moving beyond traditional design strategies. We demonstrate its capability on lightweight, high-performance metamaterials and apply it to optimize orthopaedic implants with biocompatible elastic modulus and enhanced impact energy dissipation. This approach opens new pathways for designing complex heterogeneous materials with targeted performance.
