Meta-particles are electromagnetically small scattering particles employed in the design of metasurfaces and metamaterial antennas. The process of designing meta-particles is conventionally iterative, where full-wave simulations of the meta-particle are performed while systematically changing the geometric parameters, and sometimes the material properties, of the particle. Such iterative approaches are generally resource intensive and time consuming, hence a surrogate model based approach may potentially save significant resources and time during the meta-particle design phase. In this work, we demonstrate employing two different surrogate models to estimate the scattering behavior of a “dogbone” meta-particle, namely Kriging and inverse distance. Results show that the Kriging model outperforms the inverse distance model and is capable of achieving very good approximation compared to full-wave simulation, even with a small number of design space sample points.
|