Variable-fidelity design optimization of microwave devices using multi-dimensional cauchy approximation and coarsely discretized electromagnetic models

Application of multi-dimensional Cauchy approximation and coarse-discretization electromagnetic (EM) models to surrogate-based optimization of microwave structures is discussed. Space mapping is used as an optimization engine with the surrogate model constructed as a Cauchy approximation of the coarsely discretized device EM model. The proposed approach allows us to perform computationally efficient optimization of microwave structures without using circuit-equivalent coarse models traditionally exploited by space mapping algorithms. We demonstrate our technique through design of a range of microwave devices, including filters, antennas, and transitions. Comprehensive numerical verification of the proposed methodology is carried out with satisfactory designs obtained -for all considered devices -at a computational cost corresponding to a few full-wave simulations.