面向注塑产品工艺缺陷的知识图谱构建方法及应用
Construction method and application of knowledge graph for process defect of injection molding products
葛睿夫 1任志刚 1林江豪 1林越 1高祖标1
作者信息
- 1. 广东工业大学 自动化学院粤港澳离散制造智能化联合实验室,广东广州 510006
- 折叠
摘要
针对现有注塑产品缺陷故障原因排查与定位依靠专家人工诊断效率低、成本高昂等不足,本文提出了一种面向注塑产品缺陷的知识图谱构建方法及其应用,目的在于将专家知识采用知识图谱进行表示,利用基于知识图谱的垂直检索技术,解决故障排查和定位困难的问题.首先,文章基于多源异构的故障解决方案文本构建语料库,并构建知识本体模型.其次,采用面向非结构化文本的知识抽取模型,将产品缺陷的相关专家知识从原始语料中自动抽取出来.最后,利用Neo4j图数据库实现知识存储及可视化知识图谱的构建.在所构建的知识图谱中,探索并实现了知识智能搜索、故障诊断及工艺卡解析等应用,展示了知识图谱技术在注塑领域的良好应用前景.
Abstract
In response to the deficiencies of low efficiency and high costs associated with the existing manual diagno-sis of injection molding product defects,this paper proposes a method for constructing a knowledge graph for injection molding product defects and its application.The objective is to represent expert knowledge by using a knowledge graph and utilize knowledge graph-based vertical retrieval techniques to address the difficulties in fault troubleshooting and lo-calization.Firstly,a corpus of fault resolution solution texts is built based on the multiple heterogeneous sources,and a knowledge ontology model is constructed.Secondly,a knowledge extraction model for unstructured texts is employed to automatically extract relevant expert knowledge regarding product defects from the original corpus.Finally,the Neo4j graph database is used to implement knowledge storage and the construction of a visualized knowledge graph.In the con-structed knowledge graph,applications such as intelligent knowledge search,fault diagnosis,and process card analysis are explored and implemented,demonstrating the promising application prospects of knowledge graph technology in the field of injection molding.
关键词
知识图谱/本体模型/知识抽取/专家知识/故障诊断Key words
knowledge graph/ontology model/knowledge extraction/expert knowledge/fault diagnosis引用本文复制引用
基金项目
广东省重点领域研发计划(2021B0101200005)
国家自然科学基金(62073088)
国家自然科学基金(U1911401)
出版年
2024
国家科技期刊平台
NSTL
万方数据