利基為基礎之演化式粒子群演算法應用於數值最佳化問題

A Hybrid Niching-based Evolutionary PSO for Numerical Optimization Problems

鄭進利

指導教授 ： 葉維彰

國立清華大學/工學院/工業工程與工程管理學系/碩士(2012年)

https://doi.org/10.6843/NTHU.2012.00282

摘要

粒子群優化（PSO）是一個母體為基礎的優化算法，它具有簡單性和延展性。 PSO是一個典型的全域搜索之啟發式演算法; 然而，PSO在解決方案的開採能力(exploitation)和解的多樣性方面仍然有不足之處。有鑑於此，由人工細菌遺傳算法（PBGA）的啟發下，我們提昇解的多樣性，其藉由加入PBGA中染色體突變的過程並且進一步以利基為基礎的方法做修改後來保有解的多樣性，避免在搜索過程中過早收斂。我們簡稱該算法為基於利基之混合式演化粒子群演算法（NEPSO）。我們以大量的數值函數驗證所提出的演算法，其結果顯示NEPSO在大部分的驗證函數都能保有穩健性及有效性。

關鍵字

Particle swarm optimization ； mutation ； niching ； numerical optimization

並列摘要

Particle swarm optimization (PSO) is a population-based optimization algorithm which has great potential because of its simplicity and malleability. PSO is a typical global searching heuristic, but there is still an insufficiency in PSO regarding solution exploitation and diversity. In view of this, inspired by the pseudo bacterial genetic algorithm (PBGA), we enhance the variety of solution exploitation by incorporating the PBGA process–chromosome mutation. In addition to this, a modified niching method is utilized to preserve the solution diversity, and to avoid premature convergence in search process. We call the proposed algorithm Niching-based Evolutionary PSO (NEPSO). The experimental results test several commonly used numerical benchmark functions, and show that NEPSO has very promising optimization performance.

參考文獻

[1] J. Holland, Adaptation in Natural and Artificial Systems, MIT Press, Cambridge, MA, 1992.

[4] R. Storn, K. Price, Differential evolution – a simple and efficient heuristic for global optimization over continuous spaces, Journal of Global Optimization 11 (1997) 341–359.

[5] K. Price, R. Storn, A. Lampinen, Differential Evolution a Practical Approach to Global Optimization, Springer Natural Computing Series, 2005.

[6] D. Karaboga, and B. Akay, A comparative study of Artificial Bee Colony algorithm, Applied Mathematics and Computation, vol. 214, pp. 108–132, 2009.

[7] M. Clerc and J. Kennedy, The particle swarm—explosion, stability andconver gence in a multidimensional complex space, IEEE Transactions on Evolutionary Computation, pp. 658-73, 2002.