Statistical microwave circuit optimization via a non-derivative trust region approach and space mapping surrogates

摘要

A novel yield optimization technique for microwave circuits is presented. Yield optimization of microwave circuits is obstructed by the high expense of electromagnetic simulations required in yield evaluations in addition to the absence of any gradient information. In the proposed technique, surrogates using the generalized space mapping (GSM) algorithm is incorporated with a derivative-free trust region optimization method (NEWUOA: new unconstrained optimization algorithm). Moreover, a variance reduction technique is implemented in yield estimation process. Practical examples to demonstrate this new technique are included showing its efficiency.