A LEARNING-BASED METHODOLOGY FOR DYNAMIC SCHEDULING IN DISTRIBUTED MANUFACTURING SYSTEMS

To enhance productivity in a distributed manufacturing system under he terarchical control, we develop a framework of dynamic scheduling sche me that explores routeing flexibility and handles uncertainties. We pr opose a learning-based methodology to extract scheduling knowledge for dispatching parts to machines. The proposed methodology includes thre e modules: discrete-event simulation, instance generation, and increme ntal induction. First, a sophisticated simulation module is developed to implement a dynamic scheduling scheme, to generate training example s, and to evaluate the methodology. Second, the search for training ex amples (good schedules) is successfully fulfilled by the genetic algor ithm. Finally, we propose a tolerance-based learning algorithm that do es not only acquire general scheduling rules from the training example s, but also adapts to any newly observed examples and thus facilitates knowledge modification. The experimental results show that the dynami c scheduling scheme significantly outperforms the static scheduling sc heme with a single dispatching rule in a distributed manufacturing sys tem.