適應性質群最佳化演算法應用於動態環境問題之研究

Adaptive Particle Swarm Optimizer For a Class of Dynamic Fitness Landscape

莊郁強(Yu-Chiang Chuang)

指導教授 ： 范書愷

元智大學/工程學院/工業工程與管理學系/碩士(2006年)

https://doi.org/10.6838/YZU.2006.00208

摘要

本研究主要內容為一個以最佳化為基礎的動態演算法，此演算法是用來解決一個動態反應曲面最佳化的問題。本研究提出一個適應性粒子群最佳化演算法(APSO)為了來追蹤動態環境問題中的最佳解，此問題的反應曲面是由許多圓錐所組成，而最佳解的座標為最高圓錐的頂點。所有圓椎的座標都是在初始化時隨機在可行解範圍中設定，且最高的圓錐座標會每隔一段固定的iteration重新隨機變動。為了可以在動態環境中找尋最佳解且避免粒子過分的聚集而使PSO演算法失效。APSO將所有的粒子主要分為兩個群體，分別為傳統群體與動態群體，動態群體可以持續且迅速的尋找動態環境中的最佳解。動態群體再被進一步分成數個相同數量的群體，每格固定的iteration會依序有一群粒子被重新發散，而所有的粒子都會依循PSO的速度更新機制來尋找最佳解。在本研究中APSO在動態環境中的效能與不同維度中的運作表現將被仔細探討與分析。

關鍵字

質群最佳化演算法 ； 動態環境問題

並列摘要

This thesis develops an optimization-based, dynamic algorithm for a class of dynamic fitness landscape. An adaptive particle swarm optimization (APSO) is proposed for tracking multiple peaks in a class of dynamic environment. The response surface of the dynamic environment is composed of multiple cones, and the global best solution is the position of the highest cone. The locations of cones are set randomly in initial, and the location of the highest cone will be re-randomized in each period of the iteration. In order to achieve the optimal solution in dynamic environments and prevent particles from being over-crowed, the entire population is divided into two populations. One is for the ordinary population and the other population, termed dynamic population, is designed for automatically tracking various changes quickly in dynamic system. The dynamic population is separated further into several groups, re-randomized in a period of iteration in turn and augmented with the ordinary population for searching the global optimal solution. The objective of this thesis is to evaluate the performance of the proposed adaptive particle swarm optimization on the benchmark function to achieve the optimal solution in the dynamic system.

並列關鍵字

Particle Swarm Optimization ； Dynamic Environments ； Dynamic Fitness Landscape ； Re-randomization

參考文獻

[2] Morrison, R. W., & De Jong, K. A. (1999). A Test Problem Generator for Non-Stationary Environments. Conference on the Evolutionary Computation, 3, 2047-2053.

[4] Carlisle, A., & Dozier, G. (2002). Tracking Changing Extrema with Adaptive Particle Swarm Optimizer. World Automation Congress.

[7] Parrot, D., & Li, X. (2004). A Particle Swarm Model for Tracking Multiple Peaks in Dynamic Environments using Speciation. Conference on the Evolutionary Computation, Piscataway, NJ, 98-103.

[8] Esquivel, S. C., and Coello Coello, C. A. (2006). Hybrid Particle Swarm Optimizer for a Class of Dynamic Fitness Landscape. To appear in the Engineering Optimization.

[9] Hu, X., & Eberhart, R. C. (2002). Adaptive Particle Swarm Optimization Detection and Response to Dynamic Systems. International Conference on Evolutionary Algorithms, 1666-1670.