Using Bayesian Models in Preliminary Design
Our objective is to develop a methodology to create surrogate models that enable designers to explore the interactions among design parameters at levels of detail and accuracy that correspond to the current state of design knowledge. We do not assume any specific form of the response; but we do assume that the designer is able to evaluate the system response for a set of design parameters using analysis, computation models, or physical experiments. Our goal is to create models that allow the incorporation of new information as the design progresses. As the design is refined, and new points in the design space are sampled, the scope of the model can be reduced and its accuracy increased without discarding the earlier information.