Inverse design of truss lattice materials with superior buckling resistance

Abstract

Manipulating the architecture of materials to achieve optimal combinations of properties (inverse design) has always been the dream of materials scientists and engineers. Lattices represent an efficient way to obtain lightweight yet strong materials, providing a high degree of tailorability. Despite massive research has been done on lattice architectures, the inverse design problem of complex phenomena (such as structural instability) has remained elusive. Via deep neural network and genetic algorithm, we provide a machine-learning-based approach to inverse-design non-uniformly assembled lattices. Combining basic building blocks, our approach allows us to independently control the geometry and topology of periodic and aperiodic structures. As an example, we inverse-design lattice architectures with superior buckling performance, outperforming traditional reinforced grid-like and bio-inspired lattices by ~30–90% and 10–30%, respectively. Our results provide insights into the buckling behavior of beam-based lattices, opening an avenue for possible applications in modern structures and infrastructures.

Predicting stress, strain and deformation fields in materials and structures with graph neural networks

“…5 A,B). Based on a bottom-up generation procedure featured in our previous work 57 , a dataset of 762 structures is constructed with 2 2 tessellations of randomly generated unit cells (Supplementary Materials) and split as before in training and test dataset (sensitivity analysis with training data density reported in Fig. S12 ).…”

Deep Learning in Mechanical Metamaterials: From Prediction and Generation to Inverse Design

“…Indirect inverse design employs DL models to predict the mechanical properties of existing mechanical metamaterials, followed by the use of metaheuristic algorithms, such as evolution strategy and genetic algorithms, to filter out those with the desired properties. [76,77,[198][199][200] In semi-direct inverse design, DL models map the property space to the modeling parameters, requiring the use of additional modeling processes. [73,74,109,181] In contrast, direct inverse design uses DL models to straightforwardly generate geometries that meet user-defined properties, represented by pixel images or voxel volumes.…”

“…[72] Furthermore, DL has been employed for the inverse design of mechanical metamaterials, allowing a trained ANN to generate a batch of mechanical metamaterials by taking target properties as input. [73][74][75][76][77][78] DL is a type of machine learning (ML) technique that uses ANNs with representation learning. [79,80] In DL, ANN architectures that contain multiple layers are trained using a large set of labeled or unlabeled data to perform tasks like classification and generation.…”

Machine learning-based inverse design methods considering data characteristics and design space size in materials design and manufacturing: a review

“…59 There are many other studies that performed optimization by combining the forward modeling network with optimization algorithms in various design problems. [72][73][74][75][76][77][78][79] A different approach called generative inverse design network finds the optimal designs having the desired performance by using back-propagation in neural networks. Generally, backpropagation is a process of optimizing the hyperparameters of hidden layers to minimize the loss function, the value of which quantitatively defines the error between the network's predicted result and the ground truth value.…”