一种新并行遗传算法及其应用  

A NOVEL PARALLEL GENETIC ALGORITHM AND ITS APPLICATION

作　　者：唐钟[1] 张葛祥[2] 

机构地区：[1]中国工程物理研究院计算机研究所,四川绵阳621900 [2]西南交通大学电气工程学院,四川成都610031

出　　处：《计算机应用与软件》2005年第7期9-11,71,共4页Computer Applications and Software

基　　金：国家自然科学基金资助项目(No.69574026)

摘　　要：基于量子计算的概念和原理,本文提出一种新并行量子遗传算法,即粗粒度并行量子遗传算法(CGPQGA)。该算法的核心是引入层环粗粒度并行计算模型和一种新进化策略。由于CGPQGA只需迁移搜索到的最佳个体到各个子群体,因而算法的通信开销很小。通过用CGPQGA设计控制器的应用实例表明,CGPQGA优于常规并行遗传算法,能加速子群体中最佳个体的迁移,收敛速度快,全局寻优能力强,同时具有勘探和开采的能力。Based on the concepts and principles of quantum computing,a novel parallel genetic algorithm called coarse-grained parallel quantum-inspired genetic algorithm(CGPQGA)is proposed in this paper.The key points of CGPQGA are that an extended version of coarse-grained model called hierarchical ring model and a novel evolutionary strategy are introduced.Communication overhead of CGPQGA is much less expensive because only the best individual searched is migrated to all processors.Experimental results of controllers designed using CGPQGA show that CGPQGA is superior to conventional parallel genetic algorithm and can speedup the migration of the top individuals of subpopulations and CGPQGA has the characteristics of rapid convergence,good global search capability,possessing exploration and exploitation simultaneously.

关 键 词：并行遗传算法 量子遗传算法 并行计算模型 全局寻优能力 量子计算 进化策略 通信开销 应用实例 收敛速度 粗粒度 子群体 控制器 个体 最佳 迁移 搜索 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构] O242.23[自动化与计算机技术—计算机科学与技术]