Classifier systems can reduce conceptual design cycle time

Description/Abstract

Though very widely used in preliminary and detail design, commercial parametric CAD engines have not reached their full potential yet at the conceptual stage of industrial design processes. One possible reason is their lack of robustness when it comes to generating a wide variety of geome-
tries, as demanded by the global nature of conceptual design. Such large, multi-dimensional design spaces often have regions of infeasibility, where the corresponding CAD models would lead to failure as early as the geometry generation process itself or as late as the nal stages of some expensive mul-
tidisciplinary analysis process integrated into the concept design tool. In this paper we discuss the use of Radial Basis Function classifier systems as a means of mapping out infeasible regions of the design space. The ultimate aim is to
equip the concept design tool with the ability to avoid such areas, thus saving time by reducing the number of failed simulations on unphysical or otherwise unsuitable candidate designs.