Springback optimization of deep drawing process based on FEM-ANN-PSO strategy

Abstract

Deep drawing is one of the most common sheet metal forming processes. It is largely used for mass production of parts in various shapes in automobile, packaging, and household appliance industries. During this process, the stamped part is susceptible to the process failures, especially springback if the process parameters are not correctly selected. The objective of this article is to propose a complete and an efficient optimization approach, starting from modeling and ending with identification of optimal process parameters. Our approach is based on the combination of finite element simulation, design of experiments (DOE), artificial neural network (ANN), and particle swarm optimization (PSO). Based on the comparison of simulation results with experimental results, a finite element model (FEM), which can replace the real deep drawing and predict the springback accurately, has been developed. In this article, process parameters are optimized according to their degree of importance. For this reason, the analysis of variance (ANOVA) method was used to assess the degree of importance of each of the process parameters on springback. An artificial neural network (ANN) model was developed, as a predictor, to relate critical process parameters to springback. Particle swarm optimization (PSO) is then implemented to identify the optimal values of the process parameters. The results indicate an important minimization of springback could be achieved with the use of FEM-ANN-PSO strategy. This approach can, therefore, be used for the optimization of process failures of highly non-linear mechanical systems.