A reinforcement learning based scheduling for cross enterprises collaboration in cloud manufacturing

The optimal task assignment and resource allocation strategy play important roles to form an efficient task execution and high resource utility in cross-enterprise collaboration of cloud manufacturing environment. In the cloud manufacturing pattern, users submit their tasks anytime and tasks can be finished by different enterprises anywhere. Hence, task arrival and execution processes are uncertain. How to deal with these uncertainties is one of main concerns in dynamic scheduling. In addition, the increased collaboration among different enterprises which should simultaneously considers the scheduling issue of inter-enterprises and inner-enterprises aggravates the complexity of scheduling. In this regard, we propose a two-stage scheduling model based on Q-learning (QL) to address this problem. Experimental results show that compared with other methods, the proposed algorithm performs better in terms of minimizing completion time for each dynamic arrived task as well as minimizing enterprise workload.