基于遗传算法—多智能体的FMS工件调度研究  

Research on FMS job scheduling based on genetic algorithm and multi-agent

作　　者：黄恩洲[1] 吴少雄[1] 

机构地区：[1]福建工程学院交通运输系,福建福州350108

出　　处：《广西大学学报（自然科学版）》2013年第5期1086-1091,共6页Journal of Guangxi University（Natural Science Edition）

基　　金：福建省自然科学基金项目(2009J01309);福建工程学院科研发展基金项目(GY-211057);福建省教育厅A类基金(JA11192)

摘　　要：针对柔性制造系统(FMS)一般调度方法的不足,提出基于全局黑板的多智能体调度系统,该系统建立多智能体交互过程,通过多智能体的合作快速建立调度模型,并通过优化模块对调度模型进行求解,从而获得非劣调度方案。在设计优化模块时,采用遗传算法,针对柔性制造系统调度问题的特点,改进并扩展了基于工序的编码方法,引入工序—机器的关系矩阵,从而实现解和染色体的一一对应关系,并设计算法的适值函数、选择方法、交叉和变异方法。仿真结果表明,该调度系统在求解时收敛速度快、精度较高。最后通过10个经典的柔性job-shop调度算例,与单纯使用遗传算法和禁忌搜索算法进行比较,目标值平均改善2.21%和1.04%。To overcome the shortage of general scheduling method of flexible manufacturing system （FMS）, a multi-agent system based on genetic algorithm is proposed. This is a multi-agent interac- ting procedure which can establish a temporary schedule model by cooperation of agents via overall- blackboard for a given task and can call an optimization module to solve the model. To design opti- mization module, genetic algorithm is used. According to the characteristic of flexible manufacturing system schedule problem, a special representation of solution modified through operation-based rep- resentation is proposed. Matrix of process-machine is introduced to establish solution-chromasome relations. Fitness function and selection method, crossover and mutation operator are designed. Sim- ulation result shows that the proposed system could make a faster rate of convergence and more opti- mal solution. Finally, compared with genetic algorithm and tabu algorithm, this method makes an average improvement by 2. 21% and 1.04% respectively via solving ten typical flexible job-shop schedule problems.

关 键 词：柔性制造系统 调度 遗传算法 智能体 

分 类 号：TH16[机械工程—机械制造及自动化]