Desarrollo de métodos de análisis y optimización aplicados a estructuras periódicas

Resumo

This dissertation presents a methodology that, by means of the combination of a very efficient numerical analysis technique and heuristic optimizers, allows to cope with the optimized design of planar multilayer periodical structures consisting of an arbitrary number of layers and with equal size rectangular unit cells. The use of an efficient analysis method is a basic requirement for the methodology proposed because the only relationship between the optimizers and the real problem is given in terms of cost functions involving the analysis of the structure response at each step of the optimizer. Thus, taking into account that most of the computational effort falls on the evaluation of this cost function, a multilayer CG-FFT analysis method that makes it possible to obtain the field scattered by the periodic structure in an efficient way has been developed. A thorough and detailed study concerning the efficiency of the optimization methods considered to tackle the problem has been carried out. In this study different versions of optimizers based on genetic algorithms (GA), particle swarm optimization (PSO) and simulated annealing (SA) have been considered. The methodology has been applied in the field of frequency selective surfaces (FSS) and artificial magnetic conductors (AMC) design. Moreover, several monolayer and multilayer periodic surfaces prototypes as well as an optimized AMC based low profile antenna have been built and experimentally characterized.