Deep learning-accelerated multiple design generation for sound-absorbing metaporous materials

Abstract

The design of sound-absorbing metaporous materials presents significant challenges due to the design complexity and the non-uniqueness of solutions regarding the targeted acoustic property. To address these challenges, we propose a deep learning-based approach capable of generating multiple metaporous designs that meet desired acoustic performance. Specifically, our proposed method integrates a generative adversarial network (GAN) framework into a surrogate network, guiding the generation process to ensure geometric diversity and physical consistency with the desired sound absorption coefficients. Both qualitative and quantitative evaluations confirm that the proposed model successfully generates a diverse range of unit-cell configurations that achieve the specified sound absorption behavior. Experimental results show that the proposed method outperforms existing approaches, increasing design diversity by approximately 85.8% and improving physical consistency by 64.8%. Additionally, we assess the advantages of our approach in terms of computational efficiency and design interactivity, demonstrating its capability to facilitate the exploration of metaporous designs with the desired characteristics. This study holds the potential to enhance and expedite the design process for advancing metamaterial discovery across various engineering disciplines.