Co-design Of Real-time Extension And Controller In Networked Control Systems

Networked control systems (NCSs), taking network as the transmission medium, integrate the spatial separation nodes of sensors, controllers and actors, to form a close-loop control system, which are the distributed, networked, intelligent, sharing bandwidth resources, etc. Because NCSs can effectively save the network wiring, reduce the comprehensive overhead and improve the diagnostic ability, it’s popular and has a broad application prospect and practical value. However, the traditional theory research and applications of control systems have affected by the introduction of the network, which make its design more complicated and challenging, so it has been one of the heated research topics.According to the bottom-up, this paper starts with the three aspects of the underlying network nodes, control loop, and multi-loops to complete the co-design of real-time extension and controller in NCSs, combining with the communicating, controlling and scheduling. And it builds up a self-organizing communication model of serial links and offers a method for multi-nodes to join in the network, so that devices can attain the devices’plug-and-play capability without the MAC address. The related works finish the co-design of NCSs’model and controller, which can dynamically adapt to the time-varying sampling period and solve the subprime problem caused of using global control parameters. This paper seriously proposes a two-dimensional scheduling policy of variable sampling period based on feedback bandwidth and priority based on sampling period, and it can balance the quality of control (QoC) and quality of service (QoS).The main research works and results are included as follows:Firstly, this paper proposes a self-organizing communication model based on serial links to meet the requirements on the dynamic nodes’ change from the communication aspect. It adopts the automatic scanning way to access the free communication address, and detects the address conflicts to get the new nodes added into the network efficiently. Through the physical platform’s verification, all nodes can join in the network and own the plug-and-play ability in the absence of MAC address.From the perspective of system model, this paper regards the NCSs as an uncertain dynamic range model. According to the sampling period, it uses the method of successive discretization and designs the state feedback controller, and proves the stability with the Lypapunov function and LMI. And it also uses the method of Invalid Frequency Parameter to calculate the sampling period for the one-order and two-order plants. Finally, the simulation results verify the stability of NCSs’ model and figure out the relationship of the QoC and the sampling period which does help to the next scheduling policy.In view of the network scheduling, it designs a scheduler with max bandwidth constraints and bandwidth contribution, and uses the vary-sampling period method to globally optimize the QoC and QoS. Moreover, its extra bandwidth can gradually get back to the network, so it reduces excessive consumption and guarantees enough bandwidth for new nodes to get access to the network. Then, this paper interprets the rationality of scheduler. Finally, the simulation results verify the stability and superiority of the scheduling policy from multiple perspectives.Considering the features of the vary sampling period scheduling policy and the arbitration of the Control Area Network (CAN), this paper seriously proposes a kind of priority scheduling policy based sampling period and offers a method to design the Dynamic Priority (DP) and Data Identifier (DI) separately, which reduces the overhead of dynamic scheduling algorithm particularly and owns the two CAN’s characters of interpreting the meaning of message and the link arbitration. Finally, the simulation results of the scheduling policy show the superiority of combining priority with varying sampling period scheduling algorithms, and improve the control performance of NCSs.