Interpretable AI Guides Bone-Inspired Voronoi Lattices Toward Enhanced Mechanical, Acoustic, and Vibration Response

Voronoi architectures in bone enable lightweight, efficient load bearing through remodeling under dominant loads and thus exhibit graded, load-path-aligned geometries (Wolff's law). Inspired by this architecture, we proposed aligned Voronoi lattices and further optimized them using interpretable AI to surpass empirical performance trends of conventional lattices. Relative to the baseline Voronoi lattice, which exhibits the best performance among the original geometries, the optimal Voronoi improved compressive strength, energy absorption, sound absorption, and vibration response by 94.68%, 334.77%, 76.41%, and 36.49%, respectively. These improvements were primarily attributed to the strut alignment, which suppressed premature joint failure and increased compressive load-bearing capacity by shifting the dominant load-transfer mode from bending to axial compression. In addition, reduced internal space increased viscous/thermal dissipation and enhanced broadband sound absorption. Interpretable, physically informed descriptor modeling further supported that compact, load-path-aligned struts contributed to exceptional multi-performance, and that geometric heterogeneity reduced the vibration transfer function. The mechanism-aware designs, informed by interpretable AI, can offer superior mechanical, acoustic, and dynamic performance compared with conventional lattices for engineering applications.