Application Of Predictive Control To Networked Control System

With the advantages of resource sharing, fewer cables, easy to expand and maintain, Network control system has become one of the focus in the development of automation technology. In this paper, the turbine is used as a controlled object in the network control. Aiming at the packet loss in network control system, dynamic matrix algorithm in predictive control is proposed to design controller; In view of the uncertain short time delay in the network control, this article proposes a control strategy which combines dynamic matrix control with fuzzy control. Main contributions of this article are given as follows:First, introduce the network control system in detail, analyze the difference between the network control system and traditional control system. The intervention to the network, brings some issues such as networked delay, packet loss, drive mode and packet timing disorder. This paper summarizes several major research directions in network control.Second, the basic principles of dynamic matrix algorithm and the parameters on the control performance is described. Aiming at the packet loss in network control system, dynamic matrix algorithm is proposed to design controller:In the actuator side, if control variable exists packet loss, when the time length between the time with packet loss and with no packet loss is less than predictive time domain length, the predictive control variable which the controller with no packet loss predicts is used as control variable now. when the time length between the time with packet loss and with no packet loss is more than predictive time domain length, the control variable which the actuator stores latest is used to control the controlled object in order to reduce the packet loss on control system performance; In controller side, if there is packet loss from sensor to controller, the error between prediction model and actual model before packet loss is used to correct the prediction model and reduce control performance because of feedback packet loss.Third, Hydro-turbine governing system, whose structure and adjustment purpose is recommended, is used as a controlled object in the network control. This paper has established the mathematical model of Hydro-turbine governing system in the case of small fluctuations for taking advantage of mechanism modeling method, introduced the composition, modules and usages of truetime toolbox. In the truetime network environment, the network control system of water turbine governing system is built, however, this text designs controllers utilizing dynamic matrix algorithm and PID, and compares the control effect of this two methods when existing network packet loss through simulation experiments.Finally, the uncertain short time delay in the network control is researched in this paper, which combines dynamic matrix control with fuzzy control to design network controller, in order to adapt the situation when generalized controlled object changes as the uncertain short time delay changes in the network control. The experiences adjusting dynamic matrix parameters are used for the knowledge base of fuzzy control, it sets up fuzzy rules and obtained control weighting matrix Q and Error weighting matrix R,who adapting to current network, through time delay rules and fuzzy rules, then calculate control variable by dynamic matrix algorithm. In the network control system of water turbine, this text designs controllers making use of Fuzzy Dynamic Matrix algorithm and dynamic matrix algorithm,and compares the control effect of this two methods when existing network uncertain short time delay through simulation experiments.