一种基于改进遗传算法的多目标动态调度优化  

An Improved Genetic Algorithm for Multi-objective Dynamic Scheduling Optimization

作　　者：连坤雷[1] 张超勇[1] 高亮[1] 张朝阳[1] 

机构地区：[1]华中科技大学数字制造装备与技术国家重点实验室,湖北武汉430074

出　　处：《中国制造业信息化（学术版）》2010年第2期13-17,21,共6页

基　　金：国家高技术研究发展计划(863)项目(2007AA04Z107;2007AA04Z190;2006AA04Z131);湖北省教育厅中青年项目资助(Q20092303)

摘　　要：静态调度问题一般是NP-hard问题,而动态调度问题的性能指标比静态调度的更为复杂,并且以多目标综合性能指标居多。在系统地研究了多目标动态作业车间调度问题的基础上,运用滚动窗口技术和改进的遗传算法来解决动态调度问题。滚动窗口技术将调度过程分成连续静态调度区间,在每个区间内用多目标遗传算法进行调度优化。最后开发了面向应用的动态调度原型系统,并对改进的Job-shop基准实例进行仿真试验,证明了提出策略的可行性和有效性。Dynamic scheduling problem is a more complex NP-hard problem compared with static schedulingproblem and in most cases it has multi-objective performance criteria.Based on study of the multi-objective dynamic scheduling problem,it emploies the rolling-horizon procedure and an improved genetic algoritihm.In the rolling-horizon procedure,dynamic scheduling problem is decomposed into a series of continual and static scheduling problems,multi-objective genetic algorithm is applied to each of these problems.In order to adapt to the complex manufacturing environment and sustain the stability of production,a human-compute collaborative scheduling procedure is presented for the implementation of the scheduling process.A scheduling prototype system is developed and tested on the improved job-shop benchmark instance,the simulation results validate the effectiveness of the proposed strategies.

关 键 词：作业车间调度 动态调度 遗传算法 滚动窗口 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]