Many-Objective Optimal Theory And Techniques Applied To Novel Antennas

Antennas is one of the most critical components in wireless communication systems. To relax the ever-increasing system requirements and improve the overall system performance, their designs evolve from the secondary role to topical one. A planar dual band antenna, Inverted-S Antenna (ISA), and the dielectric embedded electronically steerable multiple beam (DE-ESMB) antenna array were comprehensively and sysmetically modified and synthesized in this dissertation.The original ISA comprises a folded strip and a straight strip. Both strips create an S-shape structure. To obtain planar ISA, a coplanar waveguide is used as the feed network. However, the impedance matching is difficult in such a feeding mode, which is deviate from the physical understanding using the rule of thumb to design the original one. Thus the synthesis of ISA can only be solved by representing such an inverse problem as a multi-objective optimization one with more than three objectives, which is termed as many-objective optimization problem. Furthermore, the achievement of more compact size, broader bandwidth and omni-directional radiation pattern in azimuthal diractions requires more sophisticated designs. Therefore, the meandering technique and symmetrical geometry is proposed to modify the structure of ISA. Since no guideline is available to design this novel ISA, a many-objective optimization is inevitable.Similarly, although constructed as a combination of Yagi-Uda antennas, the synthesis of DE-ESMB excludes any analytical or empirical solutions because of the loaded dielectric, and must be formulated as an optimization problem. If all-port operation mode and equal height corner reflector is utilized to modify DE-ESMB, the angle diversity and antenna isolation can be realized. In this way, a multi-input multi-output (MIMO) antenna with exclusive features could be attainable, if the many-objective optimization technique is invoked again.Besides multiple objective functions, the optimal design problems of these antennas have numerous local optimal solutions with highly nonlinear behavior, which additionally increase the problem complexity. Evolutionary algorithms not only converge to global optimal solutions but also have the potential to produce a multiple non-dominated solutions in a single run. Therefore, they are applied extensively in multi-objective optimization community. However, existing non-dominance based multi-objective optimal algorithms have difficulties in solving many-objective problems. To handle many-objective optimal problems, several techniques have been proposed. Among them, the Multiple Single Objective Pareto Sampling (MSOPS and MSOPS-II) Algorithm is simple in implementations and has a feasible computational complexity, but has slow convergence and poor diversity of the final results. To address these problems, an improved MSOPS is proposed by incorporating a crowding operation of target vectors, a non-uniform target vector updating mechanism, and an external archive. In terms of the optimization of test functions, the synthesis of linear array and Yagi-Uda array, the proposed algorithm outperforms the MSOPS-Ⅱ and the other two well developed multi-objective evolutionary algorithms, HypE and NSGA II, in the concept of Pareto dominance. To illustrate the effectiveness of the proposed algorithm in engineering antenna synthesis problems, a circular polarized square patch microstrip antenna with truncated edge is firstly synthesized. The optimal solution outperforms the original classic design in many respects.In the synthesis of ISA and DE-ESMB, the conventional analytical technique is incapable of predicting their characteristics. The computational electromagnetics must be introduced. To this end, a methodology combining finite element method and evolutionary algorithm is proposed in this dissertation.With the help of aforementioned metodology, a planar ISA and its modified version are firstly optimized. As for the synthesis of DE-ESMB, the optimized design used in MEVIO application is conducted and compared with its traditional rival.The optimization of ISA indicates that the ISA can be only fed via folded line, that is, an ISA can be regarded as a kind of printed monopoles. The parametric analysis of the optimal design validates such assumption, and a new guideline for ISA designs is given. Moreover, thanks to the proposed meandering technique and symmetrical structure, the optimization of the modified ISA produces some optimized designs with compact size, and enhanced gain in the azimuthal plane.As far as the DE-ESMB concerned, there are two types of optimal designs. The trade-off between size and performance must be compromised. Based on the optimal designs, the physical limitation of the loaded dielectrics and the characteristics of array configurations are discussed. More loaded dielectric is recommended if the condition is permitted, and the height of monopole element is affected by such loading. In the end, a stereotype of the optimized design is fabricated, and the tested results demonstarted that the optimized stereotype outperforms its conventional rival.