Dikdörtgen şekilli mikroşerit antenlerin vekil model yöntemi ile analizi: Rezonans frekansı ve bant genişliği hesabı için modelin oluşturulması ve eniyilenmesi
Tez Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Karamanoğlu Mehmetbey Üniversitesi, Türkiye
Tez Danışmanı: Deniz Üstün, Dr. Öğr. Üyesi
Tezin Onay Tarihi: 2019
Tezin Dili: Türkçe
Özet:
Microstrip antennas have been widely used in a broad area of applications from satellite communication to cellular communication and from defend industry to biomedical applications. Therefore, the analysis and design of the microstrip antennas have become crucial research topics. Since the antennas can merely exhibit high performance around the resonant frequency as allowed by the range of bandwidth, the most important characteristics in the analysis of an antenna are the resonant frequency and the bandwidth. The advances in computational approaches such as neoromodels and metamodels lead to researchers for cheap and fast analysis of the antennas which is a challenging task. With this study, a conceptual study based on Surrogate Model (SM) for computation of the resonant frequency and the bandwidth of rectangular microstrip antenna (RMA) is presented. SM is a metamodel which imitates the input-output behavior of the system by adopting a compact and analytic strategy. This relevance is built on an interpolating function. SM is compacted by using weighting vectors in accordance with the relation between input-output dataset. Therefore, fitting the model properly is very important process for a tight SM. In order to improve the accuracy of the SM which builds on radial basis function, the weighting vectors of the SM have been to be optimized and determined by Differential Evolution (DE) algorithm. A dataset with different geometrical and electrical parameters reported elsewhere in the literature including number of 33 measured RMAs is utilized in the construction of the SM. From the total number of RMAs, 27 RMAs and the remaining 6 RMAs are used in the construction and the test of the SM respectively. Moreover, SM has been validated for the total absolute error (TAE) by making a comparison with neural network and formulation-based approaches that is presented in the literature. The results show that the SM computes the most accurate resonant frequency and the bandwidth results with the TAEs of 9.9 MHz and 0.131%, respectively.