High-fidelity verified surrogate-assisted shape–topology optimization of blast-resistant sandwich structures

Abstract

Blast-resistant sandwich structures are widely used in lightweight protective systems. Still, their shape-topology optimization remains computationally expensive because the blast response involves nonlinear transient dynamics, large deformations, contact interactions, and plastic energy dissipation. This paper presents a high-fidelity verified surrogate-assisted optimization framework for blast-resistant sandwich structures. A Gaussian-process surrogate is employed to predict the maximum back-face deflection, and an upper-confidence-bound objective is introduced to reduce the risk of selecting overly optimistic surrogate optima. The optimization considers maximum back-face deflection, structural equivalent mass, and topology regularity to obtain an efficient and interpretable load-transfer network. A joint node–rib optimization case is investigated, in which rib parameters and node-position parameters are optimized simultaneously. High-fidelity verification shows that the optimized structure reduces the maximum back-face deflection from 1.49 cm to 1.16 cm under an equal-equivalent mass comparison, corresponding to a 22.15% reduction. In a similar-deflection comparison, the optimized structure reduces the equivalent mass from 5.36 to 3.54 while maintaining a comparable deflection. The results demonstrate that uncertainty-aware, surrogate-assisted shape-topology optimization can improve both the blast resistance and the lightweight efficiency of sandwich structures.