一种动态种群不对称交叉的新型遗传算法  

Novel Dynamic Population and Anisomerous Crossover Genetic Algorithm

作　　者：张金萍[1] 刘杰[1] 李允公[1] 

机构地区：[1]东北大学机械工程与自动化学院

出　　处：《南京理工大学学报》2007年第4期444-448,共5页Journal of Nanjing University of Science and Technology

基　　金：辽宁省教育厅高等学校科学研究项目(2004D089)

摘　　要：在分析实数编码遗传算法各操作步骤的实质和不足的基础上,提出了以提高算法柔性为目的、以动态种群和不对称交叉为主要特点的新型遗传算法。在遗传寻优的每一代中,父辈个体的繁殖次数在限定的范围内随机波动,种群规模随之动态变化,依据生态平衡的原理,通过选择和复制将新一代种群规模限定于某一波动均值处。为提高新生个体的多样性及其在参数空间中的遍布性,提出并设计了不对称交叉的具体方法。针对新型算法,提出了双重选择的选择方法。经典型算例验证,所提算法具有收敛快、成功率高、抗早熟能力强的显著特点。The shortcomings and the essentials of genetic algorithm are discussed. A novel genetic algorithm is proposed whose population is dynamic and whose crossover is anisomerous. In every propagate process, the propagate times of parents random fluctuates in limitative range, and the size of population is dynamic. According to the principle of balance nature, the size of wave by the crossover and selection. In order to improve the variety tion in the parameter space, the method of anisomerous crossover is rithm, the method of double selection is presented. Calculating some other genetic algorithms, the results show that the convergence rate, the novel genetic algorthm are superior. population is kept at the average of and universality of the new populadesigned. Based on the new algotypical instants and comparing with success rate and anti-premature of

关 键 词：遗传算法 实数编码 交叉操作 寻优 收敛 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]