Research On Control And Scheduling Co-design Of Networked Control System

Networked Control System (NCS) is a synthetic application which combines control science, computer science and network technology. An NCS connects the controllers, the sensors and the actuators through one shared network to form the feedback control loop. The network resources competition and propagation delay raise a new scientific field where traditional control theory and methodology can be applied. Network scheduling, for instance, is an essential part of this novel domain. Network scheduling is a useful manners to schedule the available resource to reduce the signal collision caused by many notes accessing the network simultaneously. Control and scheduling co-design is to optimize the system performance and shared available resource. This thesis studies the popular protocols in MAC layer, network scheduling, control and scheduling co-design, optimal controller design. The main research work are as follows:Firstly, the background of scheduling, basic problems and a survey of recent advance in research areas are reviewed. Ethernet, ControlNet and DeviceNet are compared and their protocols are analyzed in detail. Further more, we also compare the real-time scheduling with network scheduling.Secondly, we present the scheduling simulation results both in MAC layer and in application layer. According to different bandwidth occupied by disturbance in MAC layer, we also analyze the state response and protocol using the scheduling diagram. Real-time scheduling methods are applied to network by the scheduler we designed, and we also compare the RM scheduling algorithm with EDF scheduling algorithm in the application layer.Thirdly, this paper research on control and scheduling co-design based on sampling period. According to the network-induced delay of the CAN, we derive the maximum sampling period through the stability condition. Then, the sampling period is optimized using the RM scheduling. The simulation proves that this method can improve the control performance. Finally, since token bus has a good quality of scheduling, we apply this characteristic to the NCS when communication resource is constrained; an optimal controller is designed to minimize the performance value within finite time. It is proved that the performance is improved.