Parameters Optimization Of Sheet Metal Drawing Process And Robust Design Of Distortional Springback

Flat stamped parts are widely used in car industry and householdappliance industry. In practical production, unqualified flat stamped partscaused by fluctuation of material parameters are difficult problems to solvein enterprises. It is an effective method combined with finite elementtechnology and robust design to solve that kind of problem in recent years.These studies were focused on simple parts and automobile cover part, theresearch about the flat stamped parts which used in household applianceindustry is very little. This kind of flat has many heterogeneousdistributions of local forming characteristics which is quite different fromthe automobile cover parts. How to get a approximation model with highprecision has been the focus of the robust design. The traditional surrogatemodel such as quadratic polynomial model, kriging model and the radialbasis function model has low prediction accuracy when they are used todescribe the process of sheet metal forming which has very strongnonlinear.To solve the above problems, the research work about how toestablish a surrogate model with higher prediction accuracy and efficientrobust optimization method funded by the Important National Science&Technology Specific Project of China was conducted as follows:In order to improve the modeling accuracy of the sheet metal formingoptimization, three kinds of mainstream approximate model had beenstudied, and the mathematical sense of Gaussian kernel width parameterhad been analyzed. A Gaussian kernel function model with multi-widthwas proposed on the basis of the radial basis function (RBF). This modelwas verified by Rosenbrock function, and the results show that the Gaussian kernel function model with multi-width has higher precisionaccuracy than the original radial basis function.An optimization method of the process of sheet metal forming wasproposed based on the Gaussian kernel function model with multi-widthand genetic algorithm. The principle of this method is as follows: throughthe orthogonal test design to determinate the influence parameters (IP), setup the objective function based on the Gaussian kernel function modelwith multi-width to, obtain the best parameters of BHF and frictioncoefficient by using genetic algorithm.Considering the parameters fluctuations, A robust design method toseek the optimal forming process with the least warping rebound of sheetmetal forming was proposed. A flat stamped parts was studied, in theoptimum combination conditions, the approximate model prediction errordown to0.04%.