Using cluster analysis for design of architectured materials
Architectured materials (mechanical metamaterials, heterogeneous materials, hybrid materials, etc., henceforth collectively referred to as "archimats"), are defined as materials whose properties are determined by their inner make-up, including composition and internal structure, not only at the nano- and micrometer-level, but also at larger length scales – up to the size of a sample or structural unit. This definition, which embraces both natural and engineered, man-made materials, has found its entrance in the mainstream materials science. The multiscale structure of archimats equips them with completely new physical and mechanical properties. A large number of studies are devoted to archimats with a high load-bearing capacity and low weight, a negative Poisson's ratio (auxetics), materials with a specific mechanical response to loading (conversion of compressive displacements to torsional ones), etc. Machine Learning techniques offer a potent platform for developing novel archimats with desired properties mentioned above. Using cluster analysis for archimats based on rod lattice structures, we will show how such archimats can be designed and studied.