Abstract: Design variables and noise factors have certain fluctuation during the process of sheet metal forming, so there exists uncertainty. Conventional optimization strategies, however, do not incorporate this uncertainty into the process of sheet metal forming, which often causes the optimal object function to go beyond constraints or the object function to become very sensitive to the fluctuation of design variables, thereby resulting in design failure. A parametric mould model is established by means of a self-developed STLMesher software. On this basis, the experimentally designed high –precision approximate model that can represent the actual forming process is combined with Monte Carlo simulation technique, thereby constituting a six sigma robust optimization design method based on the product quality engineering. The method takes the influence of various uncertainty factors into consideration in the initial stage of design, therefore it not only obtains an approximate optimal solution but also improves the reliability of design variables and the robustness of objective function. The product quality can also be improved greatly. It is the approximate model that is called in the optimization procedure, so the times of calling the finite element model can be decreased greatly and the efficiency of optimization can be improved. Numeral example indicates this method has high precision and good engineering practicability.

Key words: Experimental design, Response surface models, Sheet metal forming, Six sigma robust optimization