Simulation-driven design in microwave engineering: Application case studies

Abstract

Application of surrogate-based optimization methods to simulation-driven microwave engineering design is demonstrated. It is essential for the considered techniques that the optimization of the original high-fidelity EM-simulated model is replaced by the iterative optimization of its computationally cheap surrogate. The surrogate is updated using available high-fidelity model data to maintain its prediction capability throughout the optimization process. The surrogate model is constructed from the low-fidelity model which-depending on a particular application case-can be either an equivalent circuit or a coarsely discretized full-wave electromagnetic model. Designs satisfying performance requirements are typically obtained at the cost of just a few evaluations of the high-fidelity model. Here, several surrogate-based design optimization techniques for the use in microwave engineering are discussed. Applications of space mapping, simulation-based tuning, variable-fidelity optimization, as well as various response correction techniques are illustrated. Design examples include planar filters, antennas, and transmission line transitions structures. © 2011 Springer-Verlag Berlin Heidelberg.