Rapid optimisation of omnidirectional antennas using adaptively adjusted design specifications and kriging surrogates

A methodology for rapid design optimisation of omnidirectional antennas is presented. In order to reduce the computational cost of the geometry parameter adjustment process the authors exploit coarse discretisation electromagnetic (EM) simulations of the antenna structure under consideration as its low-fidelity model. A fast response surface approximation (RSA) model of the sampled EM data is created using kriging interpolation. The RSA model is subsequently used in the surrogate-based design process with the adaptively adjusted design specification (AADS) technique utilised as the optimisation engine. AADS allows us to account for the discrepancy between the surrogate and the high-fidelity EM antenna model. The optimised designs can be obtained at a low computational cost corresponding to a few evaluations of the high-fidelity antenna model as demonstrated through examples. Comparison with the benchmark technique indicates that using the RSA surrogates results in substantial reduction of the overall optimisation cost.