基于适应策略的迭代学习调度

ITERATIVE LEARNING SCHEDULING BASED ON ADAPTIVE POLICY

适应调度是一种基于状态 /性能反馈的定性控制方法 .适应调度知识完成从状态到调度规则的映射 ,与制造系统配置、生产任务构成、调度目标函数密切相关 ,并随调度问题的改变而改变 ,具有较强的针对性 .采用迭代学习方法 ,根据具体调度问题自动修正适应调度知识 ,提高系统运行性能 ;另一方面 ,由于引入了调度目标函数 ,使生产管理者最关心的性能得到优化 ,最大限度满足生产管理者的生产目标 .仿真实验结果验证了这一方法的有效性 .

Adaptive scheduling is a kind of qualitative control scheme based on the state performance feedback. Adaptive scheduling knowledge maps the states to a scheduling rule. It is problem specific; that is, it is closely related with the layout of manufacturing systems, the contents of production tasks and the scheduling objective functions. When used for scheduling a production task, the adaptive scheduling knowledge should be modified accordingly so as to be suitable for the task and consequently to obtain higher performance. This paper presents an iterative learning scheme that is used to refine the adaptive scheduling knowledge according to the problem scheduled. The scheduling objective, which reflects the performances of interest, is optimized during the iterative learning procedure. Experimental results demonstrate efficiency and effectiveness of the iterative learning scheduling.