Research On The Long Random Network Delay Based On Predictive Control

With the development of society, plant becomes more and more complex, the classic control systems can no longer meet the need, the biggest feature of control system in21st century is the introduction of the network structure. The future trend is that the information transfer between intelligent network node sensor, controller and actuator is through the network. Compared with the classic structure, network control systems have demonstrated a great advantage in less wiring, structural flexibility, system reliability and resource sharing. But it also brings in some problems, such as network delay and packet loss, etc. These issues make the system performance no longer good. This paper proposes the nonmodeling method and the modeling method of the long random network delay.First of all, test and analysis the network random delay, conclude the characteristics of the network delay, then model the short random network delay under four different drives.Secondly, this paper proposes a non-modeling control method for long random network delay. In this method, controller, actuator and sensor are all driven by clock, because of the network delay, the controller can not calculate well, so this paper gives a solution method and its stability analysis.Finally, a modeling control method for long random network delay is presented. Given the characters of the long rando(?) network delay, using wavelet theory, this paper decomposes the long random network delay into two parts:constant delay and short random network delay, then use the modeling method of the short random network delay, the plant and the network are modeled together, at last, predictive controller is introduced to control the generalized control object.