基于SPMD的粗粒度并行遗传算法在立体仓库路径优化中的应用  

Application of Coarse-grained Parallel Genetic Algorithm Based on SPMD in Path Optimization of Stereo Warehouse

作　　者：陈荣虎[1] 何运杰 CHEN Rong-hu;HE Yun-jie(School of Management Science & Engineering, Anhui University of Technology, Maanshan 243032,China)

机构地区：[1]安徽工业大学管理科学与工程学院,安徽马鞍山243032

出　　处：《软件导刊》2018年第12期108-112,共5页Software Guide

摘　　要：为了提高粗粒度并行遗传算法性能,缩短对立体仓库路径优化问题的求解时间,将一种单程序多数据流(简称SPMD)并行结构运用到粗粒度并行遗传算法中,并对算法进行改进。通过对自动化立体仓库拣选路径优化模型的求解,得到串行与并行计算两种情况下的运算时间与加速比,并在求解精度相差不大的情况下,将改进算法的计算时间与遗传算法、蚁群遗传算法进行比较。对比结果表明,并行计算能有效提高算法优化效率,缩短程序执行时间。该研究对于解决自动化立体仓库堆垛拣选路径优化问题有着重要的现实意义。In order to improve the performance of coarse-grained parallel genetic algorithm and shorten the time to solve the problem of path optimization of stereoscopic warehouse,this paper applies a single program multi-data stream(SPMD-based)parallel structure to coarse-grained parallel genetic algorithm and improves the algorithm.By solving the optimization model of picking path in automated warehouse,the computing time and speedup ratio are obtained in the case of serial and parallel algorithm,and the computation time of the improved algorithm are compared with the genetic algorithm and Ant colony genetic algorithm(AGA)when the accuracy of the solution is not different from each other.The results show that parallel computing can effectively improve the optimization efficiency and shorten the program execution time.This paper has a certain reference to the research of parallel genetic algorithm and also to the study of parallel computing.It is of great practical significance to solve the problem of route optimization of stacking and picking in automated stereoscopic warehouse.

关 键 词：粗粒度并行遗传算法 SPMD并行结构 自动化立体仓库 并行计算 加速比 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]