Data-oriented description of texture-dependent anisotropic material behavior

Abstract In metallurgical processes, as for example cold rolling or deep drawing of sheet metal, it is frequently observed that the crystallographic texture, and with anisotropic mechanical properties a material, evolve dynamically. Hence, to describe such necessary model functional dependence material parameters on which itself can vary locally different plastic strain histories. this work, we present new data-oriented approach parametrize yield function Barlat Yld2004-18p from micromechanical simulations textures. This accomplished by applying supervised machine learning (ML) methods express relationship between textures function. The are chosen continuously random texture one hand side, unimodal Goss Copper other. These rather common in metal forming. way, furthermore, transition isotropic plasticity severe case anisotropy be modeled, thought mimic dynamical evolution process. It found regularization strategy circumvent known non-uniqueness resulting behavior. After regularization, unique onset established, making possible train ML models excellent accuracy generalization trained able reliably predict coefficients unknown even small amount training data and, thus, correctly represent behavior various proposed method represents an efficient extension description yielding establishes way explicitly consider microstructural description, opens pathways formulate include process history.