帝國競爭演算法之多樣性機制

Imperial Competitive Algorithm with Diversity Control

蔡宇翔(Yu-Hsiang Tsai)

指導教授 ： 林志麟

元智大學/資訊學院/資訊管理學系/碩士(2012年)

https://doi.org/10.6838/YZU.2012.00095

摘要

帝國主義競爭演算法(Imperialist Competitive Algorithm, ICA)是近年發展出來的最佳化演算法，其主要概念係透過觀察人類近代歷史中，強權國家在拓廣領土、搶奪資源的競爭現象，模擬建構而產生。有許多文獻顯示此演算法在解決若干最佳化問題上有很好的表現。然而，ICA的主要問題在於常陷於區域最佳解。本研究透過探討ICA之缺陷，並針對ICA中的帝國競爭方式與殖民地移動方式重新改良。而實驗結果也顯示，改良後之ICA能夠有效提升搜尋全域最佳解的能力。

關鍵字

基因演算法 ； 粒子群演算法 ； 帝國主義競爭演算法

並列摘要

Imperialist Competition Algorithm (ICA) is a recently proposed approach for optimization problem. It is inspired by the behavior of imperialism in the mid 19th through the mid 20th centuries. Many studies have shown that ICA can achieve good results in many optimization problems. However, ICA tends to stuck in local optimal solutions. In this paper, we discuss the problems with ICA, and then propose a new algorithm that utilizes a new competition behavior among imperialists and a new moving behavior for colonies. Experimental results show that the new method effectively improves the performance of ICA.

並列關鍵字

Genetic Algorithm ； Imperialist Competitive Algorithm

參考文獻

[1] J. H. Holland, Adaptation in natural and artificial systems: an introductory analysis with applications to biology, control, and artificial intelligence: MIT Press, 1992.

[2] J. Kennedy and R. Eberhart, "Particle swarm optimization," 1995, pp. 1942-1948 vol. 4.

[3] M. Dorigo, M. Birattari, and T. Stutzle, "Ant colony optimization," Computational Intelligence Magazine, IEEE, vol. 1, pp. 28-39, 2006.

[4] S. Han, I. Muta, T. Hoshino, and T. Nakamura, "An approach of optimal design of HTS synchronous motor using genetic algorithm," Applied Superconductivity, IEEE Transactions on, vol. 14, pp. 896-899, 2004.

[5] A. Ketabi and M. J. Navardi, "Optimization of Variable-Capacitance Micromotor Using Genetic Algorithm," Microelectromechanical Systems, Journal of, vol. 20, pp. 497-504, 2011.