Simulation-driven design in microwave engineering: Methods

Abstract

Today, electromagnetic (EM) simulation is inherent in analysis and design of microwave components. Available simulation packages allow engineers to obtain accurate responses of microwave structures. In the same time the task of microwave component design can be formulated and solved as an optimization problem where the objective function is supplied by an EM solver. Unfortunately, accurate simulations may be computationally expensive; therefore, optimization approaches with the EM solver directly employed in the optimization loop may be very time consuming or even impractical. On the other hand, computationally efficient microwave designs can be realized using surrogate-based optimization. In this chapter, simulation-driven design methods for microwave engineering are described where optimization of the original model is replaced by iterative re-optimization of its surrogate, a computationally cheap low-fidelity model which, in the same time, should have reliable prediction capabilities. These optimization methods include space mapping, simulation-based tuning, variable-fidelity optimization, and various response correction techniques. © 2011 Springer-Verlag Berlin Heidelberg.