| Summary: | 碩士 === 淡江大學 === 電機工程學系碩士班 === 99 === This thesis investigates a new antenna design method by utilizing the orthogonal array (OA) for a small ultra-wideband (UWB) monopole antenna that meets the frequency band (3.1GHz~10.6GHz) of FCC requirements. By application of the OA in the design course of iterative procedure, it is found that the antenna characteristics can meet the requirements of UWB band within limited and relatively few iterations. In this study, a planar structure is tested under the enviroment of FR4 substrate with thickness of 0.8mm and relative dielectric coefficient 4.4, dielectric loss tangent 0.02. An UWB antenna is designed, simulated, and fabricated. The simulated results are verified by experimental ones. The goal is to examine a systematic approach to design an antenna structure by avoiding blind try-and -error. In this case, the objective is to quickly develop a small , light weight, low cost UWB monopole small antenna.
At first, based on the structure of the UWB monopole antenna to be examined, we can select suitable design/control parameters, and reasonable number of level for each parameter, thus an appropriate orthogonal array is established. Then through electromagnetic simulation software we can conduct the fractional factorial “experiments“ listed in the orthogonal array such that the fore-mentioned limited and relatively few number of iterations can be performed. The simulation results can then be used to calculate the fitness value of each experiment for sucessive optimization of the antenna parameters. At the end of each iteration/generation, the best canditate is updated and checked to see whether it has reached the design goal – to meet the requirements of UWB frequency band.
As compared to other methods of parameter design, the distinguished characteristic of utilizing orthogonal array for parameter design is that the number of occurrences of each parameter level is the same and/or equally-balanced for every parameter. When this method is employed to replace the full factorial experiment, it can reduce search times very effectively. In addtion to the speed up for the parameter design process, the method employed also maintain the global search for the solution space. To this end, the characteristic of uniform/balanced searching through the utilization of orthogonal array is combined with the technique of range reduction (gradual reduction of the search range iteratively) to implement the optimal search.
This thesis not only carried out the iterative optimization search for the unknown design parameters according to the response table of the design parameters at several selected frequencies, but also proposed the introduction of the random numbers (Gaussian distribution) to modify the parameters levels. By the combination of random numbers with the orthogonal array, plus multi-stage technique of range reduction, the goal is aim to make the convergence process more efficient, within limited number of iterations, and/or the convergence result is closer to the optimal solution.
|
|---|
