An extension to design of experiment for design optimization with implicit parametric models and virtual prototypes

Virtual prototyping is a new technique used by leading industry to develop new products without numerous costly prototypes. Variations to the virtual computer model can be made to identify the best overall design before the production of a few physical prototypes to validate the computation results. In this work, an adaptive response surface method (ARSM) is introduced to satisfy the needs of complex optimal designs driven by virtual prototyping. The method extends the current one-time RSM with better search efficiency and more accurate results in the design optimization of a complex problem. The adaptive response surface method applies nonlinear approximation to unknown functions, gradually improves the approximation model, and approaches the design optimum. A two-bar design example is used to demonstrate the newly introduced method.