Automated identification of dislocation structures from experimental Laue microdiffraction patterns

微劳厄衍射允许对晶体材料及其缺陷进行无损的三维微观结构分析。奈张量对理解塑性变形特别有意义，因为它提供了几何上必要的位错信息。有了这些信息，就可以估计位错结构和了解变形行为。但位错的完整和精确的三维结构几何是不可用的。目前的课题从这里开始：利用基于离散位错动力学模拟的虚拟衍射图，优化了连接衍射图与位错结构的人工神经网络。主要目标是在实验衍射模式与位错结构之间的定量统计理解。这种理解产生了一种广泛适用的工具。输入是衍射图样，输出是潜在的位错结构。该工具允许对衍射模式进行自动分析，从而能够编译当前方法无法访问的有意义的统计数据。缺陷统计为材料设计提供了对缺陷结构和行为的更好理解。

Micro-Laue diffraction allows the nondestructive three-dimensional microstructure analysis of crystalline materials and their defects. The Nye tensor is of particular interest to understand plastic deformation because it provides information about geometrically necessary dislocations. With this information one can estimate dislocation structure and understand deformation behavior. But the full and exact three-dimensional structural geometry of dislocations is not available. The current project starts here: By using virtual diffraction patterns based on discrete dislocation dynamics simulations an artificial neural network is optimized which connects diffraction patterns with dislocation structure. The main goal is a quantitative statistical understanding between experimental diffraction patterns with dislocation structures. This understanding results in a widely applicable tool. The input is a diffraction pattern, the output the underlying dislocation structure. This tool allows the automated analysis of diffraction patterns and thereby enables the compilation of meaningful statistics not accessible with current methods. Defect statistics provide an improved understanding of defect structure and behavior for materials design.