Assembly Modeling And Accuracy Analysis Method Of Complex Product Based On Digital Twin

The performance of complex product is closely related to assembly quality.And assembly error is usually taken as one of the important evaluation indicators of assembly quality.It is determined by parts' intrinsic properties such as structure error and material,also the error transmission method in the assembly process.In the manufacturing process of parts,some errors will inevitably occur,whose characterization method and application in the subsequent assembly process have always been a difficult problem in engineering.Moreover,the error transmission method is affected by various factors such as assembly technology and environment temperature.For these reasons,tradition assembly error prediction accuracy is not high.Based on the summary of the researches on methods of product assembly modeling and accuracy analysis,contents including product digital model construction,multi-clamp positioning and assembly analysis of thin-walled part,low-stress assembly optimization of bolted joint structure are deeply studied in this paper.According to the proposed method,the software system for digital twin model construction and assembly accuracy analysis of product is developed.It helps guide the machining and assembly of part,which is of great significance for improving the competitiveness of product.The dissertation is organized as follows:In chapter 1,the development history of product modeling and assembly digital technology is summarized.The shortcomings of assembly modeling analysis methods are pointed out,then the superiority of digital twin technology is brought out to improve the accuracy of product assembly analysis.Finally,the research content and organization structure of this paper are introduced.In chapter 2,a digital twin assembly model based on multilevel information is proposed.A hierarchical model of part feature is defined,and the data including discrete point cloud,registration elements,centroid position of real part is divided into three layers named geometric shape layer,pose constrained layer,and physical state layer.These three layers work together to express the complete attribute of part feature facing assembly process.Multi methods including point cloud fitting,homogeneous matrix operation,Hertz contact theory are used to realize the inverse modeling of part properties,which makes up digital twin model of real part.And digital twin frames for assembly are constructed focused on different application scenarios.In chapter 3,a statistical analysis method of digital twin rigid-flexible assembly product's form features is proposed based on hybrid vector.The deviation displacement field of batch parts's features is constructed.These features are statistically expressed from the perspectives of geometric shape layer and pose constraint layer.Covariance method is used to represent the shape correlation of parts1 scanned point cloud,and statistical rules of multi error characteristics including pose constraint translation term,pose constraint rotation term and geometric shape term are studied according to shape similarity.Vector of flexible part error is decomposed into orthogonal sub-vectors,which form the assembly vector loop combined with rigid part vectors.Then the error distribution of key feature points on product after assembly is obtained according to the discrete feature probability density distribution of the part features,which helps estimate assembly quality.In chapter 4,a method of complex product positioning and assembly accuracy prediction based on digital twin data is proposed.Fixture location sphere is adjusted based on equidistant offset surface.The effect of geometric error and physical characteristics on thin-walled part deformation is analyzed.Aiming at the minimum mean value of discrete points on thin-walled part,a multi-fixture positioning optimization method based on hybrid particle swarm optimization is proposed,to ensure thin-walled part in the positioning stage has a strong anti-deformation ability.An improving influence coefficient method is proposed to analyze the assembly error of thin-walled parts.Discrete points'errors on thin-walled parts after positioning stage are taken as the input,and true assembly error of these thin-walled parts is predicted by considering feature registration and physical interference.Then positioning and assembly operation of actual thin-wall parts are guided.In chapter 5,an accuracy assurance method of digital twin complex product for bolt low-stress assembly is proposed.Spring-stiffness method is used to describe bolted joint structure.The influence of metal plate structure parameters,process and environment factors on static equilibrium equation is analyzed and a calculation model of bolt shear stress is established.An evaluation method based on information entropy is proposed,which helps estimate the importance of input variables to single or multiple bolt stresses.Taking multiple bolt stresses as target optimization constraints,the Pareto non-inferior solution set of multi-bolt tightening torques is solved.The optimized process parameters are feeded back to workers to tighten the bolts in real time,to ensure product accuracy stability in the service stage.In chapter 6,the software system for digital twin model construction and assembly accuracy analysis of product is developed.According to the content of this paper,this system is divided into four modules,including multi-level reverse modeling of single part,statistical shape characterization and analysis of batch parts,positioning optimization and flexible assembly of thin-walled parts,and low-stress assembly optimization of bolted joint sturcture.This system verifies the feasibility of proposed methods in the paper,and has guiding significance for the molding process of actual product.In chapter 7,the main theoretical methods and innovations in this article are summarized,as well as the future researches.