Project group Data
Project group Data (D) will investigate new data-driven design methods considering the inherent multiscale setting of material composition, complex topology, and failure in mechanical metamaterials. The choice of the methods is linked closely to the available data. Well-organized material datasets collected from literature, existing materials databases, or generated from experiments and simulations in the project groups M, S, and F are required. Advanced machine learning models that enable the discovery and inversion of complex parameter-property linkages are developed whereat AM-related constraints are to be considered. Geometrical and topological measures for metamaterials will be investigated and efficient tools for their classification and exploitation are to be delivered. The descriptors will be complemented by the uncertainty of the mesoscale geometry as well as the effective properties. Uncertainty is quantified using a polymorphic formulation, e.g., based on a fuzzy-random description. The different aspects of deterministic and uncertain descriptors will be integrated with machine learning approaches to form the aspired data-driven framework for materials design and to discover process-structure-property linkages.