STEP 파일의 그래프적 표현

Abstract

Finding features in the mechanical parts can be applied to the manufacturing system that utilizes CAD. Among various kinds of CAD formats, STEP-file is a standard exchangeable format of CAD and related information. This paper proposes a transformation method for the graphical representation of STEP-file and, we can build a dataset that can use in a graph neural network (GNN). Among the STEP file contents, the instance schema that expresses the shape has 3493 features, and through this, a feature vector having 3494 elements can represent the vertices. Labeling is possible by coloring on the part, and it comprises that specific vertices marked with color are connected to the vertices representing the corresponding color. Through this study, we represent the STEP-file as a graphical representation, and we figure out the labeling method of the partial shape by coloring.