Research On Networked Control System Based On Predictive Functional Control

Networked control system (NCS) is a kind of closed loop control systems where the sensors, the controller and the actuators exchange data through network. Compared with the traditional point to point control systems, networked control systems which are framed on field bus have a lot of advantages such as high reliability, little system wiring, flexible configuration, open communication protocol, realizable information resource sharing, etc. However, the control systems communicate in serial digital line, causing nodes competing for network resources by protocol. Then some uncertainty from network environment will be brought into the control systems inevitably. The key to networked control systems analysis and design will be how to handle the issue caused by network time delay in network environment.Because of the existence of random long delay during the data transmission, the idea that the incremental predictive functional control (IPFC) strategy can be applied to the NCS is proposed. The multistep predictive strategy is employed to predict the time delay of the signal and compensate the random long time delay in the transmission, which can come to a better control result compared to the discrete PID algorithm.By using Davidon least square method as Neural Network online learning algorithm, a feed-forward compensation predictive functional self-tuning controller which is combined predictive function control algorithm based on linear model with nonlinear feed-forward gain compensation and suitable for nonlinear networked control system is constructed. On condition of the random long time delay and disturbance during the network transmission, PID algorithm can get a better control result compared with BP Neutral Network.In order to prove effectiveness of NCS algorithm design, experiment must be carried out. Therefore design and implementation of NCS hardware simulation platform is necessary. In this paper, the design of NCS hardware simulation platform uses star network structure with a switchboard connecting three computers, first as an object, second as a controller, third as network congestion Data Sources. This paper uses Matlab to write program of controller and object math model. Communication between controller and object uses WinSock API. The data exchange between VC++ and Matlab is achieved by Matlab engine. The third computer used as the resource of data blocking can realize the network time delay through the transmission of Multicast. Finally, experiment in the NCS hardware experiment platform verify the effectiveness of the theory method proposed in this paper.