本文選題：NSGA-II　切入點：多目標粒子群優(yōu)化　出處：《電子科技大學》2014年碩士論文　論文類型：學位論文

【摘要】：在實際應用中,大多數(shù)科學和工程問題都是多目標優(yōu)化問題,由于各個目標函數(shù)之間有可能是不可折衷或者相互沖突的,因此不可能有唯一確定的解,能夠使所有的目標同時達到最優(yōu),對于這些問題通常優(yōu)化得到的都是一個非支配(Pareto)最優(yōu)解集。作為最適應可重構特性的天線結構之一,可重構像素天線(reconfigurable pixel antenna)一般由若干個電小的金屬貼片陣列構成,貼片之間通過RF開關彼此連接,通過改變開關的通斷狀態(tài),能夠靈活地構造多種天線形狀,從而更易實現(xiàn)天線的可重構性能。但是,由于加載的開關數(shù)量較多,可重構像素天線的設計較為復雜,所以必須借助高效的搜索方法來挖掘天線潛在的重構能力。本文主要針對多目標智能優(yōu)化算法和可重構像素天線進行了若干相關研究,具體工作內容如下:1.提出了一種自適應的帶有精英保留策略的快速非支配遺傳算法(self-adaptive NSGA-II),通過不同特性的基準測試函數(shù)與傳統(tǒng)的帶有精英保留策略的快速非支配遺傳算法(NSGA-II)和多目標粒子群優(yōu)化算法(MOPSO)進行對比,使用收斂性度量和分布性度量指標對優(yōu)化結果進行評估,進而證明self-adaptive NSGA-II的高效性。2.使用提出的self-adaptive NSGA-II對一款方向圖可重構像素天線進行優(yōu)化,并與微遺傳算法(MGA)的優(yōu)化結果進行對比,結果表明多目標智能優(yōu)化算法在天線設計優(yōu)化中較單目標智能優(yōu)化算法具有更大的優(yōu)勢。3.在可重構像素天線中,距饋電端口距離不等的開關通斷對天線性能的影響不同。為了均衡遠近開關對天線可重構的影響,同時減少開關數(shù)量,降低天線的復雜性,提出一款非均勻尺寸像素單元的可重構像素天線,并使用self-adaptive NSGA-II對天線開關狀態(tài)進行優(yōu)化,使天線在兩個工作頻率下分別實現(xiàn)六個方向的方向圖可重構性能。
[Abstract]:In practical applications, most scientific and engineering problems are multi-objective optimization problems. All the targets can be optimized at the same time. For these problems, the optimal solution set is a non-dominated Pareto optimal solution set, which is one of the most suitable antenna structures for reconfigurable properties. Reconfigurable pixel antenna is generally composed of several small metal patch arrays, which are connected to each other by RF switches and can be flexibly constructed by changing the on-off state of the switches. Therefore, it is easier to realize the reconfigurable performance of the antenna. However, the design of the reconfigurable pixel antenna is more complicated because of the large number of loaded switches. Therefore, it is necessary to mine the potential reconstruction ability of antenna by efficient search method. In this paper, we mainly focus on multi-objective intelligent optimization algorithm and reconfigurable pixel antenna. The main work is as follows: 1. An adaptive fast non-dominated genetic algorithm with elitist retention strategy is proposed, which is self-adaptive NSGA-IIA. By using the benchmark function with different characteristics and the traditional fast non-dominance with elitist retention strategy, this paper proposes an adaptive fast non-dominated genetic algorithm with elitist reservation strategy. The transmission algorithm NSGA-II) and the multi-objective particle swarm optimization algorithm (MOPSO) are compared. The convergence metric and distribution metric are used to evaluate the optimization results, and the efficiency of self-adaptive NSGA-II is proved. 2.Using the proposed self-adaptive NSGA-II to optimize a pattern reconfigurable pixel antenna, Compared with the optimization results of microgenetic algorithm (MGA), the results show that the multi-objective intelligent optimization algorithm has more advantages than the single-objective intelligent optimization algorithm in antenna design optimization. In order to balance the effect of the distance between the far and near switches on the antenna reconfiguration, reduce the number of switches and reduce the complexity of the antenna, the switch with different distance from the feed port has different effects on the antenna performance. A reconfigurable pixel antenna with non-uniform size pixel unit is proposed, and the switching state of the antenna is optimized by using self-adaptive NSGA-II. The reconfigurable performance of the antenna can be realized in six directions at two operating frequencies.
【學位授予單位】：電子科技大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TN820;TP18

【參考文獻】

相關期刊論文 前1條

1 肖紹球,王秉中;基于微遺傳算法的微帶可重構天線設計[J];電子科技大學學報;2004年02期

，

本文編號：1646569