本研究利用螞蟻演算法(Ant Algorithm) 最佳化設計桁架結構,這方法同時考慮到桁架設計問題的尺寸(sizing)、外型(shape/configuration)與拓樸(topology)。本文所提出的方法結合了下列兩種螞蟻演算法:(i) Ant System(Dorigo et al., 1992)和 (ii) API(after Pachycondyla APIcalis)演算法(Monmarché et al., 2000)。Ant System用來解決桁架拓樸最佳化問題;API演算法則最佳化設計桁架尺寸與外型。桁架結構最佳化設計的目標是搜尋重量最輕的桁架結構,而且不違反桁架設計問題的限制條件,限制條件包括桁架連桿所能承受的最大應力(stresses in members)與桁架節點所允許的最大位移(displacements of nodes)。本文利用四個桁架問題,印證所提出桁架結構最佳化設計方法的效果,所得到結果略優於由Deb及Gulati應用基因遺傳演算法所得到的結果。
In this thesis, truss-structure optimization considering sizing, shape, and topology simultaneously is presented by using ant algorithms. The proposed algorithm consisted of two kinds of ant algorithms: (i) ant system (Dorigo et al., 1992), and (ii) API(after Pachycondyla APIcalis) algorithm (Monmarché et al., 2000). First, ant system is modified and used to optimize the topology of truss, and then API algorithm is used for sizing and shape optimization in obtained topology of truss. The goal of truss-structure optimization is to search the minimum weight of truss-structure subjected to maximum allowable stresses and displacement. To confirm the effectiveness of the proposed method, the author carry out four well known truss-structure optimization problem and compare to genetic algorithm (Deb, 2001). The results indicated that the proposed algorithm has better performance than Deb’s results.