Numerical Simulation Of Steel Aluminum Self Piercing Riveting And Prediction Of Undercut Based On GA-BP Neural Network

In recent years,with the increase of the total number of vehicles,in order to achieve energy saving and emission reduction,automotive lightweight technology has been paid more and more attention.Self piercing riveting is widely used in automobile manufacturing as a connection technology in automobile lightweight.It can realize the connection of different types of materials simply and quickly,and also has good service performance.Therefore,it is very important to study the process,analyze its forming mechanism and optimize its process parameters.In this paper,1.8 mm DP590 and 1.2 mm 6016-t4 are taken as the research objects.The influence of different process parameters on the joint geometry is studied by combining theoretical analysis,finite element simulation and experiment.The neural network prediction model of undercut is established.The main work of this paper is as follows:(1)According to the uniaxial tensile test data of the base metal,different hardening models were fitted,and the weighted combination of swift and hockett Sherby hardening model(SHS)was used to characterize the rheological behavior of the base metal;the radial compression test was carried out on the hollow cylinder part of the rivet,and the double line hardening model was selected and LS-DYNA software was used to simulate the radial compression process of the rivet Compared with the displacement curve,the parameters of the model are obtained by the method of parameter inverse.(2)The SHS hardening model of base metal and bilinear hardening model of rivet were selected,and the simulation analysis of self piercing riveting process was carried out by using simufact forming.The simulation and experimental results show that the geometry of the joint is in good agreement,and the simulation can effectively simulate the forming process of self piercing riveting.At the same time,the tension shear simulation analysis of self piercing riveting joint is carried out,and the load displacement curve shows that the simulation result considering the genetic effect of riveting forming is closer to the actual situation.(3)The length of rivet and geometric parameters of nail leg tip are determined by combining experiment and simulation.The forming process of self piercing riveting with different sheet stacking sequence and sheet strength is simulated to reveal the cause of forming cavity in the joint.The influence of depth,width and draft angle of flat bottom die on undercut is studied by single factor and orthogonal test method Note: with the increase of die depth and width,undercut decreases,while the increase of draft angle increases.The influence of die depth on undercut is the largest,followed by mold width,and the influence of draft angle is the least.The optimal die process parameters are obtained in the range of studied parameters.(4)The prediction model of undercut is established based on BP neural network,and the initial weights and thresholds of BP neural network are optimized by genetic algorithm(GA).The results show that GA-BP neural network prediction value of undercut is more accurate than BP neural network prediction value.