抄録
This paper describes a set-based design approach for achieving design robustness and flexibility based on designer's preference in the multiobjective design problem with uncertain parameters. Our approach allows designers to specify the varying degree of desirability (or satisfaction) of both a ranged set of design solutions and a ranged set of performance requirements, thus incorporating designer's preference structure and handling various sources of uncertainties. For achieving design flexibility, our approach also develops a ranged set of feasible solutions that satisfy changing sets of performance requirements. Then, a new design metric is developed to measure the level of preference and design robustness. In addition, a new uncertainty measure is developed to measure qualitative relationships (i.e., sensitivity) between design and performance variables, thus enabling design modification under uncertainty. Meanwhile, the computer-based simulation tools (e.g., FEM) are currently used overwhelmingly to generate, evaluate and validate the designed products. Therefore, this paper also presents the vision of our approach to the simulation-based design environment by adopting metamodeling techniques.
