Warm-Start Expedited Optimization of Antenna Structures Using Kriging Surrogates and Iterative Correction Scheme

摘要

Electromagnetic (EM)-driven adjustment of geometry parameters, often referred to as design closure, is nowadays a mandatory stage of antenna design process. It is executed to improve the system performance as much as possible while taking into account parameter dependencies that cannot be handled by simpler methods (e.g., theoretical models), typically employed at the initial steps of the antenna development. Unfortunately, EM-based optimization is a time consuming task, especially when the quality of the initial design is poor. This paper proposes a novel framework for accelerated antenna optimization. Our methodology involves a database of previously obtained designs, kriging interpolation models, and an iterative correction scheme for design refinement. The incorporation of pre-existing data permits generation of a good initial design as well as rapid optimization entailing the cost of just a few EM antenna analyses. The technique is demonstrated using a dual-band uniplanar dipole antenna. The structure is optimized within wide ranges of operating frequencies (2 GHz to 3 GHz for the lower band and 4 GHz to 5.5 GHz for the upper band). Benchmarking indicates that the proposed approach exhibits improved computational efficiency and reliability as compared to gradient-based warm-start optimization techniques.