本論文將會利用場型合成的概念以及演算法的應用將設計出天線陣列，第一部分會先使用拋物線方程式設計出一個波束寬為60度的寬波束天線場型，再使用最小平方法去將天線陣列的場型趨近於前面預設的理想天線場型來達到合成場型合成的目的。第二部分會先將天線陣列中各單元天線的遠場資料提出，轉移至程式中使用三種不同演算法的結合進行運算，有基因演算法、模擬退火法以及代理模型，並討論哪種演算法的結合可以有效的減少計算速度以及成功率，得出代理模型加上模擬退火法的速度會比單純使用代理模型的速度平均快46秒。第三部分會將設計好的天線陣列使用微帶線將饋電網路設計好並且實作。第四部分會介紹此方法可以達到天線陣列的任意場型，同時又擴大陣列的大小來增加天線陣列應用。;This thesis utilizes the concept of pattern synthesis and algorithmic applications to design antenna array. In the first part, a wide-beam antenna pattern with a beamwidth of 60 degrees will be designed using parabolic equations. The minimum square error method is then employed to approach the ideal antenna pattern, thus achieving the goal of synthesizing the antenna pattern.
In the second part, the far-field data from individual antenna elements in the array is extracted and processed using three different algorithms: genetic algorithm, simulated annealing, and surrogate model. The effectiveness in reducing computation time and success rate of each algorithm combination is discussed. It is found that combining a surrogate model with simulated annealing results in an average speed improvement of 46 seconds compared to using the surrogate model alone. In the third part, the designed antenna array is implemented with microstrip feed networks.
The fourth part introduces how this method can achieve arbitrary antenna patterns while simultaneously increasing the size of the array for expanded antenna array applications.
