EBSD Grain Knowledge Graph Representation Learning for Material Structure-Property Prediction

The microstructure is an essential part of materials, storing the genes materials and having a decisive influence on materials' physical chemical properties. material genetic engineering program aims to establish relationship between composition/process, organization, performance realize reverse design thereby accelerating research development new materials. However, tissue analysis methods science, such as metallographic analysis, XRD EBSD cannot directly complete quantitative structure performance. Therefore, this paper proposes novel data-knowledge-driven organization representation prediction method obtain structure-performance relationship. First, knowledge graph based constructed describe material's mesoscopic microstructure. Then learning network attention constructed, organizational input into graph-level feature embedding. Finally, embedding mapping mechanical experimental results show that our superior traditional machine vision methods.