Optimal design of high-fidelity engineering device models through space mapping

Abstract

Space Mapping (Bandler, 1993) facilitates efficient engineering optimization while avoiding direct optimization of a high fidelity model. It follows the experience and intuition of engineers. Algorithms have utilized Broyden updates, trust regions, and artificial neural networks. New developments include “implicit” space mapping, in which preassigned parameters not used in the optimization process are exploited, and “output” space mapping. Following various “cheese-cutting” illustrations, we present new results in RF, wireless and microwave circuit design, integrating electromagnetic simulations.