Abstrakt

In this thesis, optimization strategies for design of microwave passive structures including filters, couplers, antenna and impedance transformer and construction of various surroogate models utilized to fasten the design proces have been discussed. Direct and hybrid optimization methodologies including space mapping and multilevel algorithms combined with various surrogate models at different levels of fidelity have been utilized to design microwave components. The surrogate models considered included equivalent circuits, coarsely discretized models or parametrized kriging models. In all optimization strategies, outlined in this thesis, a goal function based on the locations of zeros and poles of rational approximation of circuit scattering response combined with surrogate-based optimization techniques has been utilized. In this thesis, it has been shown that the proposed strategies are applicable for complex problems where the numer of design variables is higher then ten. Moreover the proposed methods converge when the initial solution is far from the design specification. Furthermore, both magnitude and phase charasteristics are tunned in the optimization loop. Comparison with alternative techniques have been shown.

Autor (1)