Explosion Response and Multiobjective Optimization of Butterfly-Shaped Circular Auxetic Cored Sandwich Panel Based on Machine Learning

Abstract

Abstract In response to the urgent demand for blast-resistant protective structures, this study addresses two major limitations in existing research: the insufficient stiffness of conventional re-entrant auxetic honeycombs due to high porosity, and the lack of systematic multiobjective optimization frameworks that balance protection and light weighting. To overcome these challenges, a butterfly-shaped circular auxetic cored sandwich structure (BCASS) is proposed, inspired by the structural characteristics of butterfly wings. A validated finite-element model was employed to evaluate its blast resistance. Compared with conventional auxetic honeycomb sandwich structures (AHSS), the BCASS achieved up to 24.51% reduction in maximum displacement ( max D ) and 5.38% enhancement in energy absorption (EA), demonstrating superior mechanical efficiency. A comprehensive parameter analysis revealed the critical influence of geometric features on antiexplosion performance. To further optimize the design, four machine learning algorithms [random forest (RF), support vector machine (SVM), radial basis function neural network (RBFNN), and extreme gradient boosting (XGB)] were employed for predictive modeling of max D and areal specific energy absorption (ASEA). XGB and RBFNN yielded the highest accuracy for max D and ASEA, respectively, and were integrated through a stacking algorithm. Coupled with a genetic algorithm, this framework enabled multiobjective optimization, producing Pareto solutions that achieved a 13.74% reduction in ASEA at constant max D or a 13.73% reduction in max D at fixed ASEA, with a balanced solution reducing both simultaneously. Finite-element verification confirmed prediction errors below 8%, validating the robustness of the proposed machine learning (ML)-assisted optimization strategy. This work thus introduces a novel bionic auxetic core coupled with intelligent optimization, providing both theoretical guidance and practical tools for the design of next-generation blast-resistant structures.