面向代理的车间动态调度方法

Multi-agent Based Dynamic Job Shop Scheduling Method

针对车间动态调度缺乏有效方法的现状,建立了车间动态调度的数学模型.将车间组元定义为多个独立的代理(Agent),并将蚁群在觅食过程中通过释放信息素的方式进行交流与合作的行为,应用到Agent之间的协商与通信中,建立了Agent之间的协作模型,形成了基于多代理协商的动态调度方法.仿真试验对比表明,基于多代理协商的车间动态调度方法优于基于遗传算法的动态调度方法.

For the complexity of dynamic job shop scheduling problem, the previous studies focus on problem definition and developed many heuristic rules. A mathematical model of dynamic job shop scheduling problem was proposed. A novel multi-agent coordinate mechanism was established to solve this problem. The dynamic job shop is composed of several autonomous agents which coordinate with each other by simulating the ant foraging behavior of spreading pheromone, By spreading pheromone on the trails and making information available globally, ants can make optimal decisions. The test results confirm that the proposed method performs better than the genetic algorithm.