基于区间协同拍卖策略的改进多Agent车间调度模型研究

Research on Modified Multi-Agent Model for Job-Shop Scheduling Based on Interval-Synergy-Auction Scheduling Strategy

基于合同网协议的传统多Agent方法 (Traditional Multi-Agent Way,TMAW)求解作业车间调度问题,一般以单一的完工时间为优化目标,忽略了调度多目标性的要求;同时,招投标机制下Agent之间的任务优化分配本质上是单步优化,优化区间过小导致对系统整体调度性能的提升非常有限。针对上述问题,重新规划了多Agent调度模型,提出一种区间协同拍卖的动态调度策略(Interval-Synergy-Auction Scheduling Strategy,ISASS),该策略实现了一定时间窗口内时间、成本、设备负载和能耗的全局多目标优化,并可通过设置权重系数来调节优化目标倾向,很好地规避了上述传统多Agent方法的不足。最后,通过仿真测试验证了上述策略的高效性,同时研究了不同时间窗口和权重系数对调度性能的影响,可较好地指导实际生产。

In general, only time is optimized when using the traditional multi-Agent way (TMAW) to solve the jobshop scheduling problem based on contract net protocol, which ignores the requirement of multi-objective optimization for scheduling system. In essence, task optimum allocation between Agents is single-step optimization. Thus very limited boost is made to the performance of the scheduling system. To solve the problems, the multi-Agent scheduling model is replanned and an interval-synergy-auction scheduling strategy (ISASS) is proposed in this paper. Time, cost, machine load and energy consumption are optimized globally in a certain time window in this strategy. And different optimal trends can be got by changing weighting factors. Finally, a simulation test validates the effectiveness. Influence on the scheduling performance under different time windows and different weighing factors is also studied so as to guide the actual productive practices.