以擾亂式的同化作用改善帝國競爭演算法

近年來許多啟發式演算法透過模擬生物的演化而發展出來，如基因演算法 (Genetic Algorithm, GA)及粒子群最佳化演算法 (Particle Swarm Optimization, PSO)等。不同於以往的啟發式演算法，帝國競爭演算法 (Imperialist Competitive Algorithm, ICA)則是模擬各個王國之間的互動關係而提出，在許多最佳化問題上有不錯的表現，但是當問題為多維度問題時，則容易陷入問題的區域最佳解。因此本研究改良傳統ICA演算法的做法，採用調整同化方式和調整搜尋範圍的方式，並將ICA演算法混合區域搜尋 (Local Search, LS)，確實在搜尋上得到更好的最佳解。

關鍵字

啟發式演算法；粒子群最佳化演算法；帝國競爭演算法；同化作用；區域搜尋

並列摘要

Many evolutionary algorithms, like Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) that simulate natural biological activities, have been proposed in the literature. Different from the published evolutionary algorithms, Imperialist Competitive Algorithm (ICA) is a recently proposed algorithm inspired by the interactions among empires. ICA has been shown to work well in many applications, but it often leads to local optimal solutions. This study improves ICA by proposing new ways to perform assimilation actions and new adjustments to the search space and by hybridizing local searches in ICA. Experimental results show that the proposed methods often find better solutions than ICA does.

並列關鍵字

Local searches ； Evolutionary Algorithms ； Particle Swarm Optimization ； Imperialist Competitive Algorithm ； Assimilation actions

參考文獻

[23] 李旺蒼, "以粒子群最佳化為基礎之混合式全域搜尋演算法求解含凹形節線成本最小成本轉運問題之研究," 碩士論文, 土木工程研究所, 國立中央大學, 2006.

[28] 葉思緯, "應用粒子群最佳化演算法於多目標存貨分類之研究," 碩士學位, 工業工程與管理學系, 元智大學, 2003.

[29] 葉麗雯, "供應商產能有限及價格折扣下多產品多供應商最佳化採購決策 " 碩士學位, 工業工程與管理學系, 元智大學, 2003.

[25] 邱宇婷, "應用粒子群最佳化演算法於關聯法則探勘之研究," 學位論文, 工業工程與管理研究所, 臺北科技大學, 2006.

[2] B. Abdi, H. Mozafari, and A. Ayob, "Using Imperialist Competitive Algorithm to Find the Optimum Shape Design of Internally Pressurized Torispherical Dome Ends Based on Buckling Pressure," International Journal of Engineering (IJE), vol. 4, p. 387, 2010.

國際替代計量

以擾亂式的同化作用改善帝國競爭演算法

全文下載

主題瀏覽