Fast multi-objective design optimization of compact UWB matching transformers using variable-fidelity EM simulations and design space reduction

Expedited multi-objective design optimization of compact microwave/RF circuits is presented. Our approach utilizes variable-fidelity electromagnetic simulations, response surface approximations (RSAs) and design space reduction. The optimization process is conducted by means of a multi-objective evolutionary algorithm applied to the fast RSA model of the structure at hand. Selected Pareto-optimal designs are refined using output space mapping methodology. For the sake of illustration, optimization of a compact transformer matching the 50 Ohm source to the 130 Ohm load is demonstrated. Significant solution space reduction and utilization of a fast kriging RSA surrogate allows us to obtain 10-point representation of a Pareto front at a cost corresponding to only 133 high-fidelity simulations of the transformer.