Research Of Product Assembly Knowledge Recommendation Method Based On Knowledge Graph Bidirectional Fusion

Product assembly is an important part of modern manufacturing.With the development and application of manufacturing information technology,digital product assembly has become the core link of product intelligent manufacturing.Knowledge recommendation technology is the key to achieving digital product assembly.Product assembly knowledge is an instructive information resource that assembly staff continuously summarizes in product production practice.However,with the product iteration and upgrading,a large amount of multi-source heterogeneous assembly knowledge lacks an effective organization form,and it is difficult to provide assembly staff with simple and easy-to-use knowledge acquisition services.Therefore,efficient assembly knowledge organization and knowledge acquisition methods have become the key issues to improve product assembly efficiency and shorten production cycles.Aiming at the above problems,this paper studies three aspects: assembly knowledge resource analysis and knowledge extraction,assembly of knowledge graph fusion construction in assembly field,and assembly knowledge recommendation algorithm.The main research contents are as follows:The first chapter summarizes the research background and significance of this article and analyzes the relevant researches at home and abroad on the knowledge graph construction and application technology and the product recommendation technology for product assembly.The research content and organizational framework of the paper are introduced.The second chapter analyzes and classifies the knowledge resources involved in the product assembly process.A method for extracting knowledge from product assembly knowledge resources is proposed,including assembly domain entity extraction method based on statistical machine learning,step-by-step extraction method of assembly domain entity relationship based on dependency analysis,and assembly domain entity link method,realize the conversion of knowledge resources from text sentences to structured entities and relationshipsThe third chapter studies the bidirectional construction and fusion technology of product assembly domain knowledge graph.Combined with the ontology modeling method,a basic pattern layer of knowledge graph was established from top to bottom.This paper proposes a structural framework of knowledge graph expansion pattern layer,and summarizes and abstracts the concepts and hierarchical relationships in the assembly field from the bottom up.Aiming at the problem of semantic concept duplication in the bidirectional construction process,a method of product assembly knowledge semantic concept fusion was proposed,and the bidirectional fusion construction of product assembly knowledge graph was realized.Finally,the cross-slide table of a certain type of CNC machine tool is taken as an example to construct a knowledge graph in the field of product assembly.The forth chapter proposes a product assembly knowledge recommendation method based on knowledge graph and scene awareness(KG-SA).The overall framework of KG-SA based knowledge recommendation method is studied.The assembly scene factors are analyzed and defined,and a knowledge recommendation algorithm incorporating the scene factors is constructed.Knowledge graph feature learning algorithm is used to reduce the dimension of knowledge graph.Based on the recommendation algorithm that integrates assembly scene factors,an assembly knowledge recommendation algorithm based on knowledge graph and scene perception is constructed,and it is trained and solved by stochastic gradient descent method.The fifth chapter developed a product virtual assembly knowledge recommendation system.The main functional modules of the system are: virtual assembly humancomputer interaction module,knowledge graph module and knowledge recommendation module.The system has been verified in actual production assembly,which can improve the product assembly efficiency.The sixth chapter summarizes the research content and innovations of this paper,and looks forward to the content and direction of follow-up research in the field of product assembly knowledge recommendation.