柔性流水车间排产问题的一种协同进化CGA求解方法  

A co-evolution CGA solution for the flexible flow shop scheduling problem

作　　者：韩忠华[1,2,3] 朱一行 史海波[2,3] 林硕[1] 董晓婷[1] 

机构地区：[1]沈阳建筑大学信息与控制工程学院,辽宁沈阳110168 [2]中国科学院沈阳自动化研究所,辽宁沈阳110016 [3]中国科学院网络化控制系统重点实验室,辽宁沈阳110016

出　　处：《智能系统学报》2015年第4期562-568,共7页CAAI Transactions on Intelligent Systems

基　　金：中科院重点实验室开放课题资助;国家重大科技专项资助项目(2011ZX02601-005)

摘　　要：为了解决柔性流水车间排产优化问题(flexible flow shop scheduling problem,FFSP),设计了一种动态协同进化紧致遗传算法(dynamic co-evolution compact genetic algorithm,DCCGA)作为全局优化算法。DCCGA算法基于FFSP特点,构建了描述问题解空间分布的概率模型,并对标准紧致遗传算法(compact genetic algorithm,CGA)的进化机制以及个体选择方式进行了改进。在其进化过程中,2个概率模型结合最优个体继承策略协同进化,并以一定的频率进行种群基因分布信息的交流,提高了算法进化过程中的种群基因信息多样性,增强了优良进化趋势的稳定性以及算法持续进化的能力。设计实验对DCCGA算法中新引入的重要参数进行了分析和探讨,确定了最佳参数值。最后,采用不同规模的FFSP实例对DCCGA算法进行测试,与已有算法进行对比分析,验证了DCCGA算法对于解决FFSP的有效性。In order to solve the flexible flow shop scheduling problem（ FFSP）,a dynamic co-evolution compact genetic algorithm（ DCCGA） is designed as the global optimization algorithm. In DCCGA,a probabilistic model is constructed to describe the distribution of solutions of the problem,and two modifications are incorporated in the standard compact genetic algorithm（ CGA） for improving the evolutionary mechanism and individual selection method. DCCGA＇s evolutionary process is led by two probabilistic models,which contains the optimal individual inheritance strategy,and communicates with each other at a certain frequency with the population genetic information. Hence,the diversity of the population genetic information is improved during the process,and also the stability of good evolutionary trend and the capacity of continuous evolution are greatly strengthened at the same time. Moreover,the suitable parameter value is suggested based on relative experiments. And,DCCGA is measured by the benchmark problems with comparison of several effective algorithm s. The results show that DCCGA is feasible for solving FFSP.

关 键 词：双概率模型 动态协同进化 最优个体继承策略 紧致遗传算法 柔性流水车间 

分 类 号：TB497[一般工业技术]