基于聚类划分子种群的多种群遗传算法

标准遗传算法存在易于早熟,容易陷入局部最优的缺点,同时标准多种群遗传算法存在进化后期种群同质化严重的缺陷。针对这一问题,将聚类思想引入到多种群遗传算法的子种群划分中,提出了一种使用聚类方式划分子种群的多种群遗传算法,使得种群划分不再只是单纯的随机行为,而是将满足约束条件的个体根据其特征划分到不同子种群中,从而解决种群同质化问题,避免所有子种群陷入局部最优。最后,通过测试两个典型函数,验证了该算法的有效性,为多种群遗传算法提供了一种新的研究方向。

Multiple Populations Genetic Algorithm Based on Clustering Dividing Child Populations

The standard genetic algorithm has the disadvantages of easy to premature and easy to fall into local optimum,at the same time standard multiple populations genetic algorithm has serious population homogeneity in the later period. In order to resolve this problem,the idea of clustering,is introduced to the population division of the multiple populations genetic algorithm,a multiple populations genetic algorithm which uses clustering to delimit child species is proposed,which makes the population division is no longer a simple random behavior,but according to its features divides the individuals that satisfy the constraint condition into different sub populations,thus the problem of population homogeneity is resolved,all child populations are avoided to fall into local optimums. Finally,through the tests of two typical functions,the effectiveness of the algorithm is verified,at the same time a new research direction is provided for multiple populations genetic algorithm.