摘要
基于多代理系统分布式调度和集中式调度的优点,提出了将二者有机结合的系统框架,分为总体调度和多代理调度网络。前者采用遗传算法优化筛选不同的调度方案,后者则由加工资源代理和协作协议构成的代理网络,保证调度对动态制造环境的适应性。每个生产订单优先级编码为一个染色体基因,每个染色体代表一个调度方案,由多代理调度网络按照订单优先级和调度协议安排工序加工时间,然后采用提前/拖期惩罚计算染色体适值,并以此为基础优选出较优染色体,经过若干代后得到最终调度方案(订单优先级和工序时间)。最后,给出了调度实例,验证了该框架的可行性。
A system framework was set up combining advantages of distributed scheduling and centralized scheduling based on multi-agents. , The framework included a centralized part and a multi-agent scheduling net. The former adopted optimized Genetic Algorithm (GA) to select diversified scheduling schemes, and the latter consisted of resource agent and cooperation protocol to keep well adaptation to dynamic manufacturing environment. Priority of each order was coded as a gene in a chromosome with each chromosome representing a scheduling scheme. Sequence processing time was arranged by multi-agent scheduling net according to order priority and scheduling protocol. Then the fitness value was calculated based on the Early/Tardiness value of each order obtained through multi-agent scheduling net. Based on this calculation, an optimized scheduling scheme was selected. PAter generation of selections, the final scheduling scheme was achieved. An application example was provided to verify the feasibility of proposed framework.
出处
《计算机集成制造系统》
EI
CSCD
北大核心
2005年第10期1414-1418,1443,共6页
Computer Integrated Manufacturing Systems
基金
国家973计划资助项目(2002CB312200)
国家自然科学基金资助项目(60004010
60274045
60443009)
国家863/CIMS主题资助项目(2002AA11301)~~
关键词
调度系统框架
代理
遗传算法
敏捷
scheduling system framework
agent
genetic algorithm
agility