Miniaturization has become of paramount importance in the design of modern antenna systems. In particular, compact size is essential for emerging application areas such as internet of things, wearable and implantable devices, 5G technology, or medical imaging. On the other hand, reduction of physical dimensions generally has a detrimental effect on antenna performance. From the perspective of numerical optimization, miniaturization is a heavily constrained problem, with most constraints being expensive to evaluate due to involving full-wave electromagnetic (EM) simulation. A convenient way of handling such a task is a penalty function approach where constraint violations contribute, upon suitable scaling, to the primary objective, i.e., the antenna size. The penalty coefficients determining the aforementioned contributions are normally adjusted through engineering experience, which does not allow for a reliable control of antenna performance figures. This paper proposes an efficient management scheme for adaptive adjustment of penalty coefficients, which eliminates the need for objective function setup by trial and error, and ensures precise control of the design constraints. Our approach is demonstrated using three broadband antennas optimized for minimum size with acceptance thresholds imposed on the in-band matching level. The adaptive adjustment of penalty coefficients is shown to outperform experience-driven setups in terms of constraint control precision and the final design quality.
Authors
Additional information
- DOI
- Digital Object Identifier link open in new tab 10.1109/tap.2021.3111285
- Category
- Publikacja w czasopiśmie
- Type
- artykuły w czasopismach
- Language
- angielski
- Publication year
- 2022
