運用基因演算法於天線最佳化設計

本論文主要探討以基因演算法結合 IE3D 與 HFSS 兩套電磁模擬軟體，運用於最佳化天線的設計。天線設計中存在著許多變數會影響結果的效能，傳統的設計方法仍只能藉由設計者反覆地手動微調天線參數求得較佳的結果，這樣的試誤法是相當耗時的。基因演算法能針對各種困難的電磁問題，找到問題的最佳解。若將本論文基因演算法的自動化參數調整機制結合電磁模擬軟體於天線設計中，不僅可以改善天線調整的效率與材料成本的損耗，更可針對各式天線特性進行最佳化設計。研究主題主要分為三大部分。首先，使用經由測試函數成功驗證的基因演算法，針對均勻線性陣列天線與非均勻圓形陣列天線進行陣列因子最佳化，除達到較低的旁波瓣位準外，亦將陣列因子的場型與 first null bandwidth等條件納入考量。其次，利用所撰寫的基因演算法結合模擬軟體 HFSS的分析功能完成寬頻標誌天線設計，使得天線阻抗與頻寬有最佳表現，觀察此天線最佳化設計與初始之結果，可看出有顯著的變化。再者，利用基因演算法結合 HFSS進行複雜結構的碎形天線設計，除了強調天線阻抗與頻寬外，並成功在結構參數中加入角度變化，針對碎形天線的指向性進行最佳化設計，經由實作與量測成功驗證其多目標最佳化模擬結果之正確性。最後，將此項自動化技術透過圖形化介面的設計，使用者只需簡單輸入天線的結構參數和參數範圍，即可方便的進行自動最佳化設計。希望本論文研究內容與設計概念能對電磁領域帶來助益。

關鍵字

碎形天線；基因演算法；最佳化；平面天線

並列摘要

The main purpose of this thesis is to study and optimize antennas using genetic algorithm in conjunction with the IE3D and HFSS simulators. Many parameters affect the antenna performance in the design of an antenna. Traditionally, an antenna is designed using a trial-and-error approach based on the designer's electromagnetic experience. However, it is very time-consuming to tune the parameters of the antenna structure repeatedly to get a good result. The better solution of an electromagnetic problem can be obtained using genetic algorithm. Proposed automatic scheme in this thesis can improve efficiency and quality of the design works, decrease cost of materials, and increase the accuracy of the simulated results. In this thesis, the applications discussed include three topics. Firstly, we use genetic algorithm, which verified by test functions successfully, to optimize the array factor of a uniform linear and a non-uniform circular array antenna. The optimization objectives include the lower side lobe level, desired beam pattern, and first null bandwidth. Secondly, we use the designed genetic algorithm combined with the electromagnetic simulation software to optimize the logo-type antenna to achieve the reflection coefficient and impedance bandwidth. Result shows that the initial antenna performance is quite poor; however; the optimized result is much better than the initial one. Thirdly, we use the proposed method to optimize the tree-fractal antenna with the angle of the structure parameter and its directivity. Simulated and measured results of the tree-fractal antenna agree with each other very well. Obtained results show that the different complex electromagnetic problems can be solved successfully by the proposed optimization method. Finally, we use MATLAB to design the graphic user interface for the proposed method. Users can easily input the parameters and ranges of the antenna structure which is to be optimized in the optimization process. Therefore, users do not need to deal with the modification of the code. Such automatic and convenient scheme is a trend in the future. Hopefully these research works can provide great helps to electromagnetic community.