| Abstract Scope |
Generating reduced-order synthetic grain structures that accurately represent the grain structures of a material is important for efficient crystal plasticity modeling. In this study, a novel machine-learning-based approach for generating representative Euler angle datasets that mimic the crystallographic texture of EBSD datasets is introduced. The method employs K-means clustering and density-based sampling in a closed-loop iteration to create representative Euler angle datasets. Proof-of-principle experiments were performed on a quench and tempered steel. Validation of the new approach was extended to twenty datasets, including BCC, FCC, HCP, orthorhombic, and monoclinic unit cells, thereby encompassing a broad range of materials and crystal structures. Pixel-wise comparisons of the pole figures indicate a match exceeding 94%, outperforming traditional methods, which achieve less than 80% accuracy. The new method is capable of achieving a 6-fold reduction in dataset size without compromising the crystallographic texture of the EBSD data. |