Modeling, Stability Analysis, And Scheduling Of Networked Control Systems

A major trend of distributed control systems is that messages are exchanged through a serial communication network by which controllers, sensors and actuators are interconnected. This type of feedback control systems wherein the control loops are closed through a control network are called networked control systems. Networked control systems have the advantages of less wires, high reliability, increased system agility, good flexibility and share of information resources, etc.. But networked control systems have their own drawbacks, too. The medium of control network is time sharing multiplexing, so the messages can not be transmitted until the network medium is free. Given the point above, the network induced time delay is introduced into the networked control systems inevitably. Depending on the network protocol the characteristics of time delay could be constant or time-varying, which makes it hard to analyze and design networked control systems.This dissertation has analyzed the network protocols, timing components and their characteristics, and discussed the factors which affect the time delays. On the basis of that, a continuous-time model of multi-input and multi-output linear networked control systems is derived which includes several network induced time delays. On the other hand, the relation of signals in the control systems is analyzed using discrete-time control system theory. A discrete-time model of networked control systems is proposed when the controller node is either time triggered or event triggered whose coefficient matrices depend on time delays. The state-space model of the system is time-invariant when time delays are constant.From different viewpoints the stability of networked control systems is studied using time delay theory, matrix theory and linear matrix inequalities respectively. A maximum allowable transfer interval is then obtained which guarantees the stability of networked control systems. The system is asymptotically stable on condition that the actual network induced delays are smaller than the maximum allowable transfer interval. Therefore, depending on the practical application a suitable control network and a proper scheduling algorithm should be selected, in order to guarantee the actual network induced delays are smaller than the maximum allowable transfer interval. Simulation experiments illustrate the effectiveness of the presented theory.The performance of networked control systems is subject to the control loop time delays which depend on the sampling periods and the sampling instances of the control loop, so the sampling periods and the sampling instances should be properly scheduled. Using the maximum allowable transfer interval derived from stability theories a scheduling method based on the maximum allowable transfer interval is designed for a multi-loop networked control system. At last, the influence of scheduling methods on the networked induced delays is studied using a Stochastic Petri Net model.