Intelligent Design Method Of Complex Product Assembly Process Based On Knowledge Graph

The production mode of the wind turbine transmission chain has the characteristics of multiple varieties and small batches.The iterative update of its assembly process plan is frequent.At the same time,the complex assembly structure and numerous parts make the traditional assembly process design method based on manual experience with long time and great efforts,which cannot effectively reuse mass historical assembly process design cases and hidden assembly process design knowledge generated in the iterative process of products.Therefore,it is urgent to study the corresponding theories and methods to improve the efficiency of assembly process design.This paper studies the assembly process design method based on knowledge graph for complex products such as wind turbine transmission chain.The main research contents are as follows:(1)Aiming at the problem that describing assembly process information with assembly process documents and CAD files may cause inconsistent information storage form and ambiguity of design intent.Based on the knowledge graph,assembly process information model is proposed.First,the assembly process information composition of the wind turbine drive chain is analyzed.Then,four knowledge graph mode layers of assembly task,assembly operation,assembly object and assembly feature are constructed to realize the standard description of concepts and their relationships in the field of assembly process design.(2)Aiming at the huge workload of manually extracting entities and relationships from mass historical assembly files to construct assembly process knowledge graph,the entity mapping and relation construction method of assembly process knowledge graph are studied.Firstly,by parsing 3DXML file and STEP file,the mapping of assembly object entities and geometric feature entities are realized;according to the language characteristics of assembly process sentences,the named entity recognition model based on BERT and attention mechanism is proposed to realize the mapping of assembly semantic entities.Then,by parsing the fit information and interference information provided by CAD system,the assembly constraint relationship is constructed;according to the sentence structure of assembly process sentences,the dependency syntactic rules are designed to realize the construction of assembly semantic relationship.(3)The traditional assembly process design method has the poor ability of knowledge reusing and reasoning.A method of assembly process design based on knowledge graph is proposed.The similarity of the assembly or assembly unit in assembly structure,connection mode,assembly feature and fit constraint is calculated by the graph matching neural network,to realize the retrieval of reusable process items.The reasoning rules are constructed with SWRL,and the operations associated with assembly units are inferred through case information such as assembly feature type,fit constraint type and assembly unit attribute,to realize the reasoning of process items.(4)Taking the gearbox of wind turbine transmission chain as the case object,the feasibility of knowledge graph construction methods such as assembly semantic entity mapping,assembly semantic relationship construction,and assembly process design methods such as similar process item retrieval,process item reasoning is verified.The simulation experiment is designed to compared the assembly process design method of this paper with the traditional assembly process design method.The results show that the method proposed in this paper has certain theoretical and application value for improving the efficiency of the assembly process design.