Data-driven multi-objective optimization for lattice-based metamaterials

Strut-based lattice metamaterials represent a class of ultra-lightweight load-bearing and energy-absorbing materials with broad application prospects in impact protection. However, due to the vast parameter space of mesoscopic configurations and the complex nonlinear relationship between these configurations and their mechanical responses, optimizing the mechanical performance of lattice metamaterials faces significant challenges. Based on the meso-structural characteristics of strut-based lattice metamaterials, an efficient rapid digital modeling method was proposed in this manuscript. Utilizing Python scripts to drive Abaqus simulation software, we achieved batch modeling and simulation analysis of the materials. Building upon this foundation, a dataset of quasi-static compression performance for various lattice metamaterial configurations was established through finite element simulations. The reliability of the dataset was confirmed via experimental validation. Subsequently, an artificial neural network model was trained to serve as a surrogate function, which was embedded into the NSGA-II genetic algorithm to conduct multi-objective optimization design of the lattice metamaterials. The optimization yielded lattice configurations exhibiting high load-bearing capacity, superior energy absorption characteristics, and balanced load-bearing/energy-absorption performance. By integrating machine learning with numerical simulations, this work effectively reduces the computational cost of optimization design, offering technical support for the rapid performance optimization and customized design of complex lattice metamaterials.