Optimization of drawbead design in sheet metal forming based on intelligent sampling by using response surface methodology

In this study, the response surface methodology (RSM) was applied for drawbead optimization. Compared with other RSMs, the boundary and best neighbor sampling (BBNS) scheme was introduced for design of experiment (DOE) first. This kind of intelligent method can guarantee the sampling search in right direction in pre-definite interval of design variables space. For validation of developed method, the Rosenbrock function was successfully approximated by methodology; corresponding response surface (RS) appropriateness can be well predicted by analysis of variance (ANOVA). In order to improve the efficiency and accuracy of RSM, especially for nonlinear and large scale problems, a coupled BBNS strategy was suggested for drawbead optimization. In this procedure, one step forming and incremental FEM method are combined for simulation procedures. One step forming method is applied for gathering information from design space and generating initial RS model; incremental FEM method is implemented to correct the RS constructed by samples due to one step forming and build accurate RS model. Finally, a practical engineering optimization problem was implemented with suggested coupled BBNS and obtained good results.