A Closed‐Loop Framework for Inverse Design: Dynamic Training and Intelligent Optimization for Heterostructured Materials

For decades, heterostructured materials have attracted great research interest because of their remarkable mechanical and physical properties. Nevertheless, exploring the immense design space to address the intrinsic trade-off between strength and toughness through trial-and-error experiments and simulation approaches entails considerable temporal and economic expenditures. Here, we propose a comprehensive end-to-end scientific machine learning framework integrating a deep learning model (Back-Propagation Neural Network with Continual Learning, BPNN-CL) for forward prediction and an optimization algorithm (NSGA-II with Partition Monitoring and Chaotic Perturbation, NSGA-II-PMCP) for inverse optimization, and demonstrate its effectiveness using metal matrix composites as representative heterostructured materials. The forward prediction using BPNN-CL achieves errors within 10% for most experimental results, and decreases the Mean Absolute Percentage Error of elastic modulus' testing dataset more than 14.9% compared with six conventional machine learning methods. In contrast to conventional approaches (e.g. NSGA-II, Bayesian optimization, et al.), NSGA-II-PMCP enables inverse design with enhanced generalization and high efficiency, and maintaining broad applicability across diverse material systems and properties. This work establishes a unified platform for heterostructured material design that, when integrated with emerging cutting-edge manufacturing techniques such as 3D printing, offers a readily extensible approach to accelerate the development of diverse advanced heterostructured materials.