Computationally efficient simulation-driven antenna design using coarse-discretization electromagnetic models

摘要

Electromagnetic (EM) simulation and simulation-driven optimization are ubiquitous in contemporary antenna design. Unfortunately, accurate simulation is CPU intensive so that employing EM solvers directly in the optimization loop may be impractical. Here, several computationally efficient simulation-driven design techniques are considered for antenna design. All of these methods exploit coarse-discretization EM antenna models, which, after suitable correction, serve as reliable prediction tools guiding the optimization process. As the coarse-discretization model is computationally cheap, time costs of the design can be greatly reduced. The approaches presented here include multi-fidelity optimization, adaptive design specifications, and space mapping with kriging-based coarse models. Application examples are given.