Robust optimization of stamping process based on Bayesian estimation

Robust designs have been widely used in process optimisation to improve product quality, which is affected by many parameters, including material properties, geometric conditions, and forming processes. In this article, a new robust tolerance model with a nested optimisation structure that considers parameter uncertainty is proposed. Bayesian estimation based on the genetic algorithm method is also presented to quantify the metamodel uncertainty. To improve the optimisation efficiency, a grey relational analysis method is employed for selecting the key controllable and uncontrollable parameters. A Land Rover stamping part from NUMISHEET 2016 is analysed, and the proposed robust tolerance model is applied. The optimisation results are compared with those based on the robust tolerance model with a single loop optimisation structure. The results indicate that the design approach of the proposed tolerance model, considering the parameter and metamodel uncertainty, can effectively reduce the springback and improve the quality of stamping parts.