进化计算中多物种多目标算法优化研究

摘要

进化算法是智能计算中涉及到组合优化问题的一个子域,模仿种群活动、自然规律等通过优胜劣汰的自然选择极值和遗传信息的传递,不断迭代演化寻求问题的最优解.各种算法的相互影响和促进使进化计算得到了飞速发展,关于多物种协同进化的研究也逐渐深入.针对大多文献中多物种的基因型和表现型映射关系一致,不能体现多物种的不同基因型的问题.提出了一种基于矩阵映射的多物种不同基因型构造方法,可以构造出不同基因型的多物种.并利用矩阵数学模型分析了不同基因型多物种构造的原理和形成机制.还分析了构造不同基因型映射矩阵的性质,并给出了映射矩阵修改的方法,可以给相关研究不同基因型多物种多目标优化算法提供一些借鉴.%Evolutionary algorithm is a branch of intelligent computing,which involves combinatorial optimization.Various evolutionary algorithms imitate population activity or the law of nature,and constantly iterate to seek the optimal solution of problem by natural selection and genetic operator.Various algorithms influence and promote between each other,so that evolutionary computation has been developed rapidly.The study on multi-species co-evolution moves towards increasing sophistication,but genotype and phenotype of multi-species have the same mapping relation.This same mapping relation can not reflect different genes of multi-species.A method of generating different genotype species based on matrix mapping was proposed and analyzed in detail in this paper.By using matrix mathematical model,the principle and formation mechanism of different genotype multi-species were given in this paper.In this paper,the nature of the mapping matrix was analyzesd,and the method of mapping matrix modification was given.We also gane some suggestions for the optimization research on multi-species and multi-objective algorithm.