Machine learning – enabled inverse design of shell-based lattice metamaterials with optimal sound and energy absorption

Abstract

Currently, the development in shell-based lattice, is increasingly focused on multifunctionality, with growing interest in combining sound and energy absorption. However, few studies have explored the multi-objective inverse design process. Herein, we propose a new approach using machine learning (ML) to optimise both the mechanical and acoustic performances of shell-based lattices. Firstly, the K-Nearest Neighbour and Artificial Neural Network are employed to predict the properties of different configurations. Then the non-dominated sorting genetic algorithm is employed to generate the desired structures. Finally, the lightweight metamaterials generated achieve optimal multifunctional performances (an energy absorption capacity of 50% higher than typical Gyroid structure and a sound absorption coefficient near 1 at specific frequency band). Besides, the potential trade-off phenomenon of mechanical and acoustic properties is also presented by our work. Overall, this work presents a new concept to use ML and genetic algorithm for multi-functional inverse design for shell lattice metamaterials.

Keywords

• Lattice metamaterial
• machine learning
• non-dominated sorting genetic algorithm
• energy absorption
• sound absorption