Abstract: The product models of existing companies are mainly stored in a neutral geometry model file format.However,the neutral geometry model lacks assembly feature semantics crucial for guiding the actual assembly process,as it primarily encapsulates geometric semantics like points,lines and surfaces.To standardize geometric semantics in the neutral geometric model,a neutral geometric semantic model named Knowledge Graph-based Neutral Geometric Semantics Model (KG-NGSM) based on the knowledge graph was proposed.The geometric semantic extraction method for the neutral geometric model was proposed to extract geometric semantic elements and map them to entities in the data layer.The similarity calculations using both the knowledge graph and graph matching network was conducted to reconstruct assembly feature semantics within the neutral geometric model.Furthermore,a system for reconstructing assembly feature semantics was developed to verify the feasibility of semantic reconstruction of assembly features for wind turbine products.The research result showed that the proposed method effectively guided the actual assembly operation of wind turbine products.

Key words: neutral geometry model, geometric semantics, assembly feature semantics, knowledge graph, similarity calculation

摘要： 现有企业的产品模型主要以中性几何模型文件格式储存。由于中性几何模型是以点、线、面等几何语义形式来储存信息,在实际装配过程中,中性几何模型不具有装配特征语义,不能够为装配环节提供指导作用。针对这一问题,首先提出了基于知识图谱的中性几何语义模型KG-NGSM,规范化地表示中性几何模型中的几何语义;提出一种面向中性几何模型的几何语义抽取方法,抽取几何语义元素并映射为数据层中的实体;基于知识图谱与图匹配网络进行相似度计算,从而实现中性几何模型装配特征语义的重建;最后,以某企业风力发电机产品为例,研发了装配特征语义构建系统并进行验证,结果表明,所提方法对于风机产品的实际装配操作环节具有一定指导作用。

关键词: 中性几何模型, 几何语义, 装配特征语义, 知识图谱, 相似度计算