An Explicit Model Predictive Control Of Networked Control Systems

With the fast development of network technologies and the extension of the control system scale, the structure of control system is changing. The network and communication technology are applied to the control field and lead to the new development of Networked Control System, which are distributed, networked and intelligent. And which have the advantages of less wires, good flexibility, high reliability and share of information resources, etc. However, since the communication network has been introduced in the networked control system, some new issues such as time delay, data dropout and network scheduling appear.The key issues of time delay and data dropout in the networked control system, and the problem that the traditional model predictive control, which should optimize online repeatedly, is difficult to adapted to the high-frequent sampling or real-time systems, this paper presents a low complexity explicit model predictive control algorithm to handle the time delay and data dropout in the network control system. The algorithm can off-line obtain the corresponding partition of each state explicitly for state feedback optimal control law. The on-line computation effort is only restricted to a table-lookup. It prohibits the computational complexity of on-line optimization, and improves the speed of on-line computing, and meets the higher requirements of real-time networked control system. Main achievements are as follows:1. Based on studying a great deal of related literature, the detail summary of some basic contents, and recent research on networked control system is made, and the basic principle of explicit predictive control is introduced.2. A networked control systems simulation software-TrueTime is introduced. Using the TrueTime toolbox, two control algorithms (fuzzy-pid and PID) and scheduling algorithms (EDF and RM) are simulated and analyzed. 3. As to the time delay in network control system, This paper presents an explicit model predictive control. Based on multi-parametric quadratic programming method, the algorithm obtains an explicit control law on state feedback. And based on this control law, the piecewise affine model is established. Since the algorithm does not need to optimize online repeatedly, it improves the on-line calculation speed of the controller, The algorithm deals with the random communication time delay well. Simulation results show the effectiveness of the method.4. To the existing network-induced data dropout, As the model for networked control systems with data packet dropout is a switching system, we transform it into piecewise affine form. We can switch the PWA model to simulate the process of packet dropout of network. then we use the principle of explicit model predictive control to solve a constrained optimal control problem with PWA model. We get a best control law which is piecewise explicit to the states based on multi-parametric quadratic programming method.Through the application of this control law, we get the networked control systems with data dropout, which is based on the explicit predictive control. Simulation results show the effectiveness of the method.