摘要
借鉴生物学中的非达尔文进化理论,并针对传统遗传算法中可能存在的"早熟收敛"、种群多样性维护困难等方面的问题,提出一种无传统"选择"操作的"中性进化"算法模型.该模型基于木村资生中性进化思想,侧重交叉和变异纯随机遗传操作,并采用自适应种群更新策略,使得算法结构更为简洁、多样性维持能力有所加强.若干数值类型的多峰函数测试以及复杂经典NP类318城市货郎担问题实验研究,已初步显示出中性进化算法在抑制"遗传支配"现象、避免陷入"早熟收敛"以及改进遗传算法的全局优化效果方面与Holland传统遗传算法相比性能较优的特性.
Considering the arguments in biology non-Darwin＇s evolution theory and the existing problems of traditional genetic algo- rithm, that was easy to fall into the premature convergence and was difficult to maintain population diversity. This paper proposed a kind of neutral evolutionary algorithm model without traditional selection operation. This model was based on the evolutionary theory of Kimura Motoo and focused on the pure random genetic operation of crossover and mutation. It use adaptive update strategy. This make structure of model uncomplicated and make population diversity maintenance better. Through the experiment of some multimodal test functions and the classic NP 318 cities the traveling salesman problem, some experiment results have shown that neutral evolution- ary algorithm model is superior to traditional Holland genetic algorithm model in restraining ＂genetic dominance＂, avoiding ＂prema- ture convergence＂ , and improving global optimization effectiveness etc.
出处
《小型微型计算机系统》
CSCD
北大核心
2014年第4期824-826,共3页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(70071043)资助
关键词
遗传算法
中性演化
早熟收敛
TSP问题
genetic algorithms
neutral evolutionary
premature convergence
traveling salesman problem
