Sampling Techniques for Sequential Kriging Metamodels in Robust Design Optimisation

摘要

The goal of robust design optimisation is to improve the quality of a product or process by minimising the deteriorating effects of variable or uncertain parameters. This robustness can be achieved by different approaches using formulations of statistical decision theory which all require many function evaluations throughout the optimisation. In cases where no closed-form descriptions of the objective are available and Point-wise solutions are expensive to evaluate, metamodelling techniques based on design of experiments are used to replace the actual numerical analysis codes. In this paper we propose a method to augment the significance of the metamodel in the vicinity of crucial points through additional sampling of selected points.