Design of Lattice Structures With Customized Mechanical Response via Forward–Inverse Machine Learning

Lattice structure are mechanical metamaterials with excellent specific strength and specific energy absorption (SEA), which have received extensive attention in recent years. However, due to the intricate relationship between structure and mechanical response, customized design of lattice structures with unconventional mechanical response remains a challenge. This research proposes a forward–inverse machine learning framework based on artificial neural network (ANN) and adaptive improved nondominated sorting genetic algorithm (NSGA‐II), referred to as NSGA‐II–ANN, for designing plate lattice structures with customized mechanical responses. The dataset consists of geometric features of random lattice structures, efficiently and accurately captured by a digitalization method, and mechanical responses obtained from finite element simulations. The trained NSGA‐II–ANN exhibits the ability to accurately predict the mechanical response and SEA of various lattice structures, and can also perform inverse design of the structures. Its reliability is verified by finite element simulations. In addition, the effects of plates with different positions and geometries on the mechanical responses are quantitatively analyzed. In summary, distinct from conventional trial‐and‐error and empirically driven design paradigms, the trained NSGA‐II–ANN framework eliminates the need for complicated solid mechanics calculations, establishing an efficient and versatile approach for development of customized lattice structures.