Design and reinforcement-learning optimization of re-entrant cellular metamaterials

Abstract

The demand for cellular metamaterials exhibiting multiple desired properties has become increasingly prominent due to the complexity of engineering applications. In this study, a novel dual-functional re-entrant cellular metamaterial is proposed for excellent bandgap characteristics and enhanced energy absorption capacities. A structural evolutionary route of the metamaterial unit cell is developed through the introduction of flexural ligaments and geometric circles, which leads to the achievements in both superior bandgap and enhanced energy absorption. Firstly, the wave propagation characteristics of cellular metamaterials in three evolved configurations are analyzed systematically. Bandgap properties and the generation mechanism are revealed by mode shape analysis. Then, the Q-learning algorithm in reinforcement learning is employed to optimize significant structural parameters of cellular metamaterials to acquire the maximum bandgaps. The certain stability and efficiency of the algorithm are discussed by the evolutionary optimization of metamaterial unit cells with different configurations. Additionally, the energy absorption capacities of metamaterials with optimal microstructure configurations are investigated numerically. Plateau stress and specific absorption energy are compared under various impact velocities, with improved performance observed in the novel cellular metamaterials. The findings of this study offer a promising avenue for advancing the development of dual-functional metamaterials with tailored properties for diverse applications.