AI-assisted symmetry-informed topology optimization of woven materials for broadband sound absorption

Woven materials are promising candidates for lightweight broadband sound absorption, yet their rational design is hindered by the high computational cost of acoustic simulations and the combinatorial complexity of discrete weave topologies. These challenges limit the effectiveness of conventional topology optimization and random design exploration. Here, we present a symmetry-informed, surrogate-based approach for the inverse design of woven acoustic materials. Physically valid patterns are generated using a constraint-aware parametric–topological model that enforces crystallographic wallpaper symmetries and manufacturing feasibility. High-dimensional binary topologies are embedded into a compact latent space through a semi-supervised convolutional autoencoder trained on both labeled acoustic data and unlabeled feasible patterns. Gaussian process regression is then employed to predict acoustic performance and guide efficient global optimization. The proposed approach achieves accurate reconstruction and reliable acoustic prediction, shows independent FEM re-evaluation errors below 5% for the optimized designs, and identifies optimized woven architectures with enhanced broadband absorption. Beyond performance improvement, the analysis reveals systematic structure–function relationships, demonstrating how symmetry operations and fundamental domain size govern transport properties and dissipation mechanisms. These findings establish a general foundation for symmetry-informed design of woven acoustic materials and provide a scalable pathway toward data-driven optimization of complex textile and metamaterial architectures. Woven materials hold potential for lightweight broadband sound absorption, but their design is limited by computational and combinatorial challenges. Here, the authors introduce a symmetry-informed, surrogate-based approach using a parametric-topological model and machine learning to optimize woven acoustic materials, achieving enhanced absorption and uncovering key structure-function relationships for scalable design.