Repozytorium publikacji - Politechnika Gdańska

Manufacturing tolerances and other types of uncertainties may considerably affect operation and performance of microwave components and systems. Quantification of these effects is therefore an important part of the design process. It is even more important to obtain designs whose sensitivity to parameter deviations is reduced as much as possible. All of these require statistical analysis carried out at the level of electromagnetic (EM) simulation models, which may be prohibitively expensive when executed directly, e.g., through Monte Carlo simulations. In this work, a technique for fast yield optimization of compact microstrip couplers is proposed. Our methodology involves variable-fidelity EM models, local surrogates constructed using a response feature approach, as well as a sequential approximation optimization framework. A combination of these tools permits yield optimization at a remarkably low cost corresponding to less than a dozen of high-fidelity EM analyses of the circuit under design. The proposed approach is demonstrated using a miniaturized equal-split rat-rate coupler operating at 1 GHz. Reliability of results is validated using direct (high-fidelity) Monte Carlo simulations of the initial and optimized designs.

Autorzy

• prof. dr inż. Sławomir Kozieł link otwiera się w nowej karcie ,
• dr hab. inż. Adrian Bekasiewicz link otwiera się w nowej karcie