一种新型遗传算法的研究  

Analysis of a New Kind of Genetic Algorithm

作　　者：杨震[1] 敖发良[1] 

机构地区：[1]桂林电子工业学院通信与信息工程系,桂林541004

出　　处：《数据采集与处理》2003年第4期444-449,共6页Journal of Data Acquisition and Processing

摘　　要：根据自然界普遍存在的进化收敛现象 ,本文提出了一种新型的混合遗传算法 ( Mix genetic algorithm,MGA)。MGA算法将生物的进化过程建模成一个在表现型与基因型两个层面上同时进行的生命过程。MGA不仅考虑了表现型层面上的进化过程 ,还引入了基因型层面上的进化及两个层面间的相互映射关系。本文通过二维 Rosenbrock函数数值优化的例子 ,展示了 MGA良好的在线性能指标和离线性能指标 ,并分析了 MAG算法实际上是“r-K策略连续统”。理论分析表明 ,MGA算法总能以概率 1收敛于最优解。MGA算法性能优良 ,容易编程实现 ,并在一定程度上克服了基本遗传算法搜索时的盲目性 ,有较好的应用前景。A hybrid genetic algorithm (mix genetic algorithms,MGA) is proposed based on the evolution convergence existed in the world. The evolution takes place both in the phenotype and the genotype. MGA algorithm considers a evolution of the phenotype, and the evolution of the genotype and the mapping between them. An example of Rosenbrock function optimization is given. The applicational example shows a result both in on line performance and off line performance. And MGA shows that it is a r K continuum of strategiest.The further analysis shows that MGA can globally converge with probability one. MGA is easy to be implemented and can overcome the blindness in GA′s parameter configuration to some extents. The method has highly practical significance. Finally simulational results are given.

关 键 词：遗传算法 进化收敛现象 随机化搜索算法 混合遗传算法 

分 类 号：O242.23[理学—计算数学]