论文部分内容阅读
差分进化算法简单、高效且鲁棒性好.然而在求解大规模优化问题时,其性能随着问题维度的增加会迅速降低.针对此问题,提出一种基于MapReduce编程模型的分布式差分进化算法.算法采用改进的精英学习策略和岛模型两种机制,提高算法的收敛精度.利用MapReduce并行编程模型,构建分布式差分进化算法,并将其部署到分布式集群Hadoop上.利用13个标准测试问题进行仿真实验,实验结果表明该算法求解精度高,且具有较好的加速比和扩展性,是求解大规模优化问题的有效方法.
Differential evolution algorithm is simple, efficient and robust.However, when solving large-scale optimization problems, its performance will decrease rapidly with the increase of the problem dimension.Aiming at this problem, a distributed differential evolution algorithm based on MapReduce programming model The algorithm uses improved elite learning strategy and island model two mechanisms to improve the accuracy of the algorithm.Using MapReduce parallel programming model to build distributed differential evolution algorithm and deploy it to the distributed cluster Hadoop.Three standard tests The experimental results show that the proposed algorithm has high solution accuracy and good speedup and expansibility, which is an effective method to solve large-scale optimization problems.