Optimization Of Sheet Metal Forming Process Parameter Based On Support Vector Machine Regression Algorithm

Complicated nonlinear problems like geometric nonlinearity, materialnonlinearity and boundary nonlinearity are generally involved in sheet metal forming.Unreasonable design variables will result in forming defects like wrinkling, crackingor scrap parts in severe cases. With the rapid development of computer science andoptimization theory, optimization of stamping process parameters based on the finiteelement analysis and optimization algorithm has become a research focus in the field.The complicated variable relation in the process optimization of sheet metalforming is studied in this paper to improve the accuracy of optimization model andobtain the best process parameters. Some details investigated in this paper are statedas follows.(1) The common approximate model technologies and experimental designmethods are discussed, and the basic theory of Support Vector Machine (SVM)regression is stated in detail.(2) Standard Particle Swarm Optimization(PSO) algorithm is improved by usingthree kinds of strategies, nonlinear dynamic improving inertial weights, adjustingaccelerated factor and adopting adaptive mutation particles. The accurate supportvector regression model between process parameters and forming quality isconstructed using Support Vector Machine. Based on the model, the Improved ParticleSwarm Optimization (IPSO) algorithm is applied to search the optimizationparameters of stamping process.(3) With an example of box-shaped deep drawing work-piece, the principle ofselecting design variables and sheet metal forming quality evaluation index are stated,also an optimization model is established. It’s proved with the test part that the fittingaccuracy of the nonlinear function constructed by SVM has a better advantage overResponse Surface Method, neural networks and Kriging model. Based on the SVMmodel, the IPSO algorithm is applied to search the optimization model.(4) The SVM and IPSO algorithm are applied to the sheet shape optimization ofauto machine cover plate and the process parameters optimization of floor beams, theresults indicate that this method is widely effective and practical in the stampingoptimization.