Aplicación de algoritmos genéticos y recocido simulado a la reconstrucción del diagrama de radiación de antenas

Abstract

The performance of different local and global optimisers applied to antenna far-field pattern prediction from planar near-field information is presented. The method proposed replaces the antenna under test by equivalent magnetic surface currents whose components are computed from reference near-field data using an appropriate optimisation technique. From the optimised equivalent model, antenna far-fields can be reconstructed. The Nelder Mead downhill simplex algorithm, along with global optimisers such as genetic algorithms (GA) and simulated annealing (SA) were all investigated and compared. Numerical results demonstrate that hybrid GA outperform local optimisers, classical real and binary encoding GA and SA. Results of near-field to far-field transformation are also included and discussed.