并行列队竞争算法及应用  

Parallel Line-up Competition Algorithm and Its Application

作　　者：史彬[1] 张和烽[1] 鄢烈祥[1] 

机构地区：[1]武汉理工大学化学工程学院,武汉430070

出　　处：《武汉理工大学学报》2010年第20期69-73,共5页Journal of Wuhan University of Technology

基　　金：国家自然科学基金(20976142)

摘　　要：提出了一种求解大规模复杂优化问题的并行列队竞争算法。该算法采用带虚拟主节点的粗粒度模型作为并行框架,通过动态迁移拓扑、动态繁殖子代等策略以实现局部搜索与全局搜索的对立统一。对典型测试函数的计算证实:并行列队竞争算法的求解性能与收敛速率明显优于普通列队竞争算法,适用于求解大规模复杂优化问题。This paper presents a parallel line-up competition algorithm（PLCA） for solving large-scale complex optimization problems.A coarse-grained parallel model with virtual master node has been presented to implement the parallelization.A new dynamic migration topology and dynamic offspring reproduction scheme are employed to balance well global search and local search.Comparative research on a group of benchmark functions show that PLCA is superior to conventional line-up competition algorithm,and it is suitable for large-scale complex optimization problems.

关 键 词：并行计算 列队竞争算法 粗粒度模型 动态环形拓扑 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]