基于改进遗传算法的工作流调度研究  

Research of Workflow Scheduling Based on Improved Genetic Algorithm

作　　者：王晓军[1] 熊潇[1] 

机构地区：[1]南京邮电大学信息网络技术研究所,江苏南京210003

出　　处：《计算机技术与发展》2013年第7期108-111,115,共5页Computer Technology and Development

基　　金：国家科技支撑计划(2007BAH17B04)

摘　　要：在资源受限工作流系统中,任务的执行顺序和资源分配对工作流执行时间有很大影响。文中就此问题提出了一种新的方法,把工作流调度分为计划和执行两个阶段,先运用改进的遗传算法对工作流系统中的任务执行顺序和资源分配做好全局优化,然后再按照计划执行,达到执行时间最短的目的。实验结果表明,与动态工作流调度方法以及标准遗传算法比较,在相同工作流中,当存在并发执行的任务时,基于改进后的遗传算法的调度方法能够做到全局资源分配最优,使得整个工作流系统在执行时间方面最短。In the workflow system of limited resources, task execution sequence and resource allocation has a great effect on execute time of workflow. Present a new method, workflow scheduling is divided into two parts are plan and execute. First, use improved genetic algo- rithm in workflow scheduling to optimize task execution sequence and resource allocation globally, and then perform as planed to achieve the purpose of shortest execution time. Finally, compare improved genetic algorithm with dynamic workflow scheduling method and canonical genetic algorithm in the same workflow. The results show that when exist concurrent tasks, scheduling method based on improved genetic algorithm can optimize resource allocation globally and make the entire workflow system achieve execution time shortest.

关 键 词：工作流调度 资源分配 遗传算法 精英保留 动态变异 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]