<p>In this study, a novel design method for prestressed metal-ceramic interpenetrating structures is proposed. The aim is to enhance the internal prestress level and improve anti-penetration performance through structural optimization, thereby improving structures with excellent protective performance. First, numerical calculation methods were used to compare the prestress distribution characteristics of different types of metal-ceramic interpenetrating structures. We also correlated structural parameters with prestress levels to construct a surrogate model for rapid prediction and efficient performance evaluation. We then used a combination of the non-dominated sorting genetic algorithm (NSGA-II) and the surrogate model to optimally design the parameters and types of prestressed metal-ceramic interpenetrating structures.</p><p>The results show that the NSGA-II approach can obtain a better Pareto solution set than traditional gradient-based optimization methods, significantly improving optimization efficiency and effectiveness. Furthermore, the anti-penetration improvement provided by prestress in typical structures was enhanced by the optimization. This further verifies the positive effect of the optimized design on the anti-penetration capability of the structure. The combined optimization strategy, based on a backpropagation neural network (BPNN) surrogate model and NSGA-II algorithm, proves adaptable and feasible for improving structural performance. This study provides an effective method and theoretical support for the high-performance structural design of prestressed metal-ceramic interpenetrating composite materials and offers a reference for the engineering application of complex prestressed ceramic protective structures.</p>