并行遗传算法研究综述  

A review of parallel genetic algorithms

作　　者：高家全[1] 何桂霞[1] 

机构地区：[1]浙江工业大学之江学院,浙江杭州310024

出　　处：《浙江工业大学学报》2007年第1期56-59,72,共5页Journal of Zhejiang University of Technology

基　　金：浙江省教育厅基金(20051436)

摘　　要：扼要概括和评价了并行遗传算法的四种并行模型:主从式模型、粗粒度模型、细粒度模型及混合模型,不仅给出了每种模型的特点及适用范围,而且也指出了每模型在应用中需解决的关键问题.通过模型比较发现粗粒度模型以其较小的通信开销和对种群多样化,获得了最广泛的应用.最后介绍了目前并行遗传算法的主要评价模型,并指出了并行遗传算法的今后主要研究方向,将向理论化、模型化、异步化及混合化等方向发展.Four kinds of parallel models of parallel genetic algorithms, such as master-slave model, coarse-grained model, fine-grained model and hybrid model, are simply generalized and evaluated. For every model, its characteristics and applicable fields are displayed. Then some key problems which need be solved for improving its application are suggested. Through comparing these models, it is shown that coarse-grained model is more widely applied for its little communication overhead and its diversity of population than other models. At the end, some main evaluation models of parallel genetic algorithms are presented. Furthermore, it is shown that researches of parallel genetic algorithms will focus on theorization, modelization, asynchronization etc in the future.

关 键 词：并行遗传算法 并行计算 并行模型 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]