Research and Implementation of a Real-time Task Dynamic Scheduling Model based on Reinforcement Learning

The traditional task scheduling method is insufficient in queuing delay, system load balancing and CPU utilization.It is important to build a real-time task scheduling model for the high efficiency and real-time nature of scheduling.In this paper, a real-time task dynamic scheduling model based on reinforcement learning is proposed for the problem slower of task deadline and low communication in the existing task scheduling method of intelligent production line. The model divides scheduling into two stages: task sorting and task assignment, aiming at minimizing queuing time in the task sorting stage, and using dynamic Reinforcement learning method (DRL) to get the task sequence with the shortest queue time. During the task assignment phase, more accurate scheduling decisions are made by analyzing the load of the processor. The simulation results show that the model adjusts the task queue in real time under the condition of changing the task data, which is obviously better than the earliest deadline priority algorithm in task sorting performance. The load balance of the system is better than the method of random task allocation, which improves the utilization rate of CPU and the quality of service of the system, and effectively reduces the missrate of task deadline in the task scheduling process.