Please use this identifier to cite or link to this item:
Investigations and Improvements on Particle Swarm Optimization for Global Optimizations
|Contributors: ||NTOU:Department of Systems Engineering and Naval Architecture|
Particle swarm optimization;Neighborhood particle swarm optimization;Taguchi method;Shape optimization
|Issue Date: ||2011-06-28T08:19:57Z
|Abstract: ||摘要:近年，仿生物群聚智慧技術(Bio-mimetic swarm intelligence)開發及應用於複雜系統處理問題，已受到國際學者重視，於2003年規劃為一新學術領域－群聚智慧。其中，粒子群演算法是最廣受各領域重視的方法之一；由Eberhart與Kennedy在1995年提出，係仿自然界中鳥群遷移補食之行為，於搜尋域中，每粒子決定移動之方向，除了参考本身的經驗外，同時也受到群體中最佳粒子的影響。 本計劃依粒子群演算法的處理流程，提出系統式探討粒子群演算法於各類型問題之適性：1）首先，探討参數值的適當取用：採用田口法以及参數關係數學式，透過一系列之數值實驗測試，將其結果予以歸納之；2）於局部區域之搜尋性能：選擇數學規劃法中之常用兩種直接搜尋法比對檢討；3）仿網路拓樸鄰域型粒子群演算法之概念，提出以之鄰近區域爲粒子更新参考模式之擾動型粒子群演算法；4）最後，將此演算法應用於結構工程設計問題，並檢討其可行性。|
abstract:Recently, the developments of Bio-mimetic swarm intelligence and theirs applicationsto complicated systems are to be increasingly noticed by scholarships all over world. As a result, in 2003, a new academic field involved these techniques is established. Particle swarm optimization, one of them, had been proposed by Eberhart and Kennedy, in 1995. The particle swarm optimization (PSO) is inspired from the moving scheme of immigration and foraging for bird school. The moving direction of one particle in swarm over space is determined by the position of cognition’s experienced and these of the best particle. In this research, according to the procedure of PSO, we propose a systematic study onthe searching characteristics of PSO for various types of optimization problems. Four stages included in the systematic study are: 1) Firstly, using Taguchi method and mathematic relationship expression of parameters in PSO is to find their suitable values through a series of experimentaltests. 2) Second, to examine and discuss the performance of local search by comparison of results from PSO and two classical direct methods. 3) Third, using the concepts of the network topology of neighborhood PSO, a perturbed PSO, based on neighbor regions of the best experienced position, , of particle i and the best position of swarm,, is proposed and to examine its performance through several benchmark problems. 4) Finally, thePSO is applied to study shape optimization of hatch corner of container-ships and several engineeringdesign optimizations with constraints.
|Appears in Collections:||[系統工程暨造船學系] 研究計畫|
Files in This Item:
All items in NTOUR are protected by copyright, with all rights reserved.