摘要
针对传统分布估计算法局部搜索能力弱,易陷入早熟收敛的问题,在分布估计算法的基础上引入精英策略并采用划分子种群独立进化的方式,提出一种基于精英协同的多种群分布估计算法。该算法混合了两种后代产生的策略:一种是进化过程采用精英协同操作用于进行局部搜索并开辟出新的搜索空间,另一种是采用划分子种群独立进化方式保证种群间个体的多样性。基准测试函数实验结果表明,该算法在收敛性和多样性方面均表现出明显优势。
Aiming at the problem of weak ability of local search and falling into premature convergence easily which exist in the traditional estimation of distribution algorithm( EDA),a new algorithm based on the distributed estimation algorithm is proposed to solve the problems. In this algorithm,the elite strategy is introduced and the method of dividing the sub populations is adopted. The algorithm combines two types of reproducing strategies,one is using the elite cooperative operation for local search the evolution process and developing new searching areas,the other is dividing population into sub ones to ensure the diversity by independent evolution. Experimental results of benchmark test functions show that the algorithm has obvious advantages in both convergence and diversity.
出处
《计算机应用与软件》
2017年第1期281-285,共5页
Computer Applications and Software
关键词
分布估计算法
早熟收敛
精英协同
子种群
Estimation of distribution algorithm
Premature convergence
Elite cooperation
Sub population