Springback Prediction Research And Experiment On Two-axle Flexible Roll Bending

The two-axle flexible roll bending technology had obvious advantages,such as realizing dieless forming,manufacturing workpiece straight edge,and the high surface quality which can improve the fatigue life of it obviously,The springback is one of the main factors that affect the forming precision of flexible roll forming.The method of finite element simulation and neural network model is used to predict springback.The main works of the paper were as follows:1.Basing on the elastic-plastic theory and hertz theory,the process of loading forming and unloading springback of sheet metal were analyzed in the two-axle flexible roll bending technology.And then,the Elastic-plastic theory model was established suitable for the two-axle flexible bending process.2.The influence of the diameter of the rigid roller,the material parameters and the thickness of the plate on the springback of the workpiece was analyzed.For the 2024-T351 aluminum alloy,the relationship between the thickness and the forming limit dimension is non-linear.The larger the thickness is,the smaller the forming limit size is.There is a linear relationship between the limit press-in and the thickness.3.By using ABAQUS,a three-dimensional numerical simulation model of two-axis flexible roll bending is established.The flexible roll bending simulation of 20 # steel and 2024-T351 aluminum alloy with thickness of 1.5mm is carried out.Compared with the experimental data,the error is 5% ~ 8%,and the consequent increase in the amount of press-fitting,the error decreases.The stress distribution of the surface of the flexible roller is simulated by finite element simulation.The change of the diameter of the rigid roller with the press-in amount is simulated and the force variation of the rigid roller with different diameters is summarized,and the mathematical expression between them is obtained,which can provide reference for machine design.4.By using genetic algorithm to optimize the initial weight of BP neural network and establishing the prediction model of two-axis flexible roll bending neural network,the advantages of BP neural network and genetic algorithm make the model have good forecasting and global searching ability,and the experimental data as input data.The results show that the output data and the experimental data of the neural network optimized by genetic algorithm have higher precision.4.The characteristics of this paper is that the flexible roll bending technology of sheet metal is used to analyze the influence of multi-process parameters.The theoretical analysis,experimental design,FEM and neural network are applied to the springback prediction of flexible roll bending.Under the precondition of guaranteeing the analysis precision,the design parameter of the process parameter is improved,which has practical significance to the development of precision forming technology.