Fast, Robust and Global Optimisation for Antenna Design using Meta Modelling

This paper introduces a data-driven meta-modelling framework designed to optimize challenging antenna designs, addressing the limitations of traditional gradient-based - and global search methods. The framework utilizes Bayesian Op-timization (BO) and High-Order Gaussian Processes (HOGPs) to approximate black-box functions, substantially reducing the reliance on full simulations. Two case studies that 1) optimises a multi-section corrugated horn antenna and 2) balances gain and return loss in a dual-reflector system, demonstrate the framework's effectiveness in handling complex design challenges, offering a scalable and efficient tool for antenna engineers.