Design Of Sliding Mode Predictive Controllers And Performance Optimization For Networked Control Systems

With the increasing of the size of control systems, the network will play more and more important role in the control systems. There is a need for infiltrating the techonology of communication and network in the industry control. The networked control system is a new research domain in the automation. The research on the networked control systems is of important academic and applicable values.NCS has the following outstanding advantages: reducing the installation and maintenance fee, enforcing the modularity, flexibility, expansibility and reliability of the system, realizing resource share and long-range manipulation and control, having the good ability of diagnosis, exchanging the information in modules(sensor, controller, actuator) of the system through network.Because the network has been introduced in the control systems, some new problems appear, traditional control theory including stability has to be researched again, the following problem is inevitable:(1) The delay of transferring the information is often random. Furthermore, the delay is often greater than a sampling period.(2) Because of the undependability of network transfer, there is a dropout in NCS.(3) Because of the limit of network bandwidth and data-packet capacity, the multiple channel transfer is evitable.(4) In the case, it is impossible that all the sensor/actuator can use the network at the same time, which decides that NCS needs a scheduling algorithm except a control algorithm to determine which sensor/actuator can send/receive the data through network. Now a pure control problem becomes a coupling of control and schedule problem.Considering the above, NCS is researched in this thesis. The sliding predictive controller with good performance is designed for a rapid-changed plant under delay, network induced noise and data dropout circumstances. At the same time, how to optimize the performance of NCS and how to integrate the MATLAB algorithm in the NCS software are considered. The main contents and contributions are asfollows:(1) The detail summary of the research actuality of NCS is made.(2) The slide mode predictive controller is designed for NCS, which can deal with long time delay, overcome the matching condition perturbation and some dropout.(3) The experiment is designed for the measurement of the network-induced delay based on Ethernet. Through the experiment, the delay characteristic is got.(4) To reduce the communication amount of NCS, the plant information is used in the controller node. The simulation shows that the sampling period guaranteed the system is stable is greatly enlarged, which shows that the proposed method helps to reducing the communication burden in NCS essentially.(5) In order to utilize the MATLAB algorithm in the NCS software, a new method of integrating the MATLAB algorithm in VC++ through DLL is introduced, and illustrative examples are given.