Research On Stability Analysis And Scheduling Methods For Networked Control Systems

With the rapid development of control technology, network technology and computer applied technology, networked control systems(NCSs) has received more and more attention, and it is becoming an important issue in automation field. The NCSs have the advantages of information sharing, much fewer wire, easy reconfiguration, easy maintenance, high efficiency, high reliability and high mobility. However, because of the introduction of network in closed-loop system, there are many questions should be further studied, such as the stability problem and network scheduling. The dissertation is concerned with the basic problems in NCSs, the modeling, control and scheduling of NCSs are studied, and the main contents are as follow:For a class of NCSs which only the sensor and controller are connected with network, and the states of the plant are unmeasured, a state observer is designed to compensate for time delay and data dropout. Then by using two different kind of modeling method:one is discrete-time modeling, the other is discrete-time and continuous-time modeling, the whole closed system is described by a maths model, at last, the stability of NCSs are analyzed, and the simulation result is given, which shows the advantage of the discrete-time and continuous-time modeling method.The NCSs which sensor, controller and actuator are connected with network is analyzed further. When the sensor, controller and actuator are different drive modes, by analysis the characteristic of NCSs in different nodes drive mode, proper modeling method given before are chosen. Then based on data packet dropout rate, the whole closed system including the state observer is modeled as asynchronous dynamical system constrained by configuration event rate. Sufficient and necessary conditions for exponential stability of the closed loop system are obtained, and simulation example is done in truetime for comparison and validation.For the class of NCSs given above, when the time delay is stochastic, the stochastic time delay is regarded as an incertitude factor. Then the whole system is modeled as an uncertain linear discrete system, Sufficient and necessary condition for robust H∞stability of NCSs is obtained. Then focusing on both time delay and data packet dropout, the data packet dropout process is considered as a stochastic process, by using Lyapunov principle, robust H∞stability of the system is analyzed, sufficient and necessary condition for robust is obtained. Then based on the exponential stability analysis method and robust H∞stability analysis method, the robust H∞stability with exponential convergence of the closed loop system is analyzed, simulation result shows the advantage of this method.Control and scheduling are two important research topics of NCSs. As the control loops need coordinating, the scheduling methods are analyzed for the NCSs with many control loops after the analysis for only one control loops. A scheduling method on network-induced error and LQG (Continuous linear-quadratic-Gaussian) theory is designed to solve collision of communication in networked control systems. For under scheduling, by monitoring network-induced error, a scheduling method of changing sensor sample periods and to compensate sample data in controller is put forward. At last, the simulation validates the effectiveness of the proposed scheduling method.