Research And Design Of Networked Control Systems

The networked control systems (NCS) connects various network nodes that are distributed in different locations via limited network resources to form a closed loop control system. Compared with traditional point-to-point control system, the NCS has a variety of advantages such as better information sharing, less system wiring, and easier to extend and maintain. At the same time, scale expansion, structure complication, data sharing and remote control are the demands of modern large-scale industrial process control system, the NCS can just meet the needs because of the advantages mentioned above. However, as the network used as communication medium in the NCS has uncertainty, not only the control strategy and control equipment but also the current network status can affect the control performance. Research on control system based on network environment emerges as the requirements of actual control system development and draws more and more attention of the scientists and engineering technicians. Study on the NCS not only possesses important theoretical value, but also possesses important practical significance.In this paper, system modeling, control algorithm, experiment platform and engineering application based on the NCS are discussed. The main contents are as follows:Firstly, the research background and significance of the NCS are discussed, the research status of control strategy of some kinds of NCS are reviewed and the problems need to be solved are pointed out. After that, some network problems that practical control system is facing are studied, and the network environment is monitored relying on actual project. Finally, the discrete mathematical model of the controlled object considering the network status is given based on different drive mode of different nodes. Then the model error caused by the network uncertainty is obtained using the model. All the work lays the foundation for the follow-up study.In view of the problem of random packet loss both exists in forward and feedback channel, a dual-channel compensation strategy is put forward based on the generalized predictive control algorithm. First, the compensation queues of control quantity and object output are designed respectively based on the predictive information. Then the operation mode of compensation strategy related to current network status is given. Finally, the optimization algorithm on the basis of multi-step predictive value of control quantity is designed. Simulation results show the effectiveness of the algorithm.Considering that the network environment may affect the precision of the controlled object model, a neural network predictive control strategy combining with genetic algorithm parameter optimization is proposed. Two BP neural networks which are used respectively for model prediction and control quantity calculation are designed. In order to suppress the influence of network uncertainty on the control effect, the neural network weights are modified according to the state of the controlled object real-timely. Besides, the best initial parameters of the control system are optimized by the genetic algorithm in order to reduce the difficulty of parameter tuning. Simulation results prove the effectiveness of this strategy.In view of the problem of random time delay (including short time delay and long time delay) of NCS, A compensation strategy which is called clonal immune predictive algorithm is proposed. First, discrete mathematical model of the object considering the network environment is given based on driving mode of the network nodes. Then, the short time delay is compensated using neural network according to model error and the long time delay is compensated using predictive compensation queue respectively. Simulation results prove the superiority and effectiveness of this strategy.The demands of practical industrial process control system such as high safety, reliability and stability make it difficult to directly apply existing theory results of NCS into production practice. Aiming at this problem, an experiment platform of NCS is designed and developed. First, the new control strategy can be conformed and validated by real network environment or the virtual network established. Then the implementations of some key technology of the platform such as drive modes of network nodes, communication protocols, delay measurements and clock synchronization is given. Besides, the methods of configuration, parameter setting and the operating principle of the control system are introduced. At the same time, the integration scheme of control strategy and experiment platform is designed. Finally, an experiment using clone immune predictive algorithm is carried out on this platform in order to validate the effectiveness of the control strategy and the practicality of the platform.