Research Of Adaptive Control In Network Control Systems

Network control systems are systems whose sensors, actuators and control units are connected through communication networks. This type of system has the advantage of greater flexibility with respected to traditional control systems. It can not only reduce wiring and installations cost, but also permit greater agility in diagnosis and maintenance procedures. Consequently, the networked control systems have a broad application prospects. On the other hand, few disadvantages also appear, specifically network induced delay, data packet dropout, single-packet or multiple-packet transmission, network scheduling and so on. Therefore, the research of networked control systems brings the new opportunity and the challenge for control domain. In this dissertation, the research focus on the problem of the network induced delay and dropping packet in the networked control systems. The main task is as follow:Firstly, NCS's development process is reviewed, basic concepts and basic problems are introduced, current situation of study is summarized. Then propose the main work the thesis done.Secondly, a model reference adaptive control scheme for NCS with network induced delay is developed. A Smith predictor is used to reduce the effect of network induced delay, simplify the system structure and facilitate the next design of adaptive algorithm. The controller of MRAC can not only guarantee the stability of the closed-loop system but also preserve a given level of performance when circumstance changes or a fault occurs. Finally, simulation results are included to show the efficiency of the proposed scheme.Then, an adaptive predictive control scheme is proposed for networked control systems with model error of controlled plant, data packer dropout and network induced delay. The main idea is using the capacity of the transmitted data packet sufficiently, controller obtain the input-output data through proper approach, model parameters then can be identified based on the received data. The identified model is used to design predictive controller, and the derivation of the algorithm. Simulation shows that the proposed control scheme can make system output trace the design output precisely when there exists network induced delay, data packet dropout and model errors.Finally, it summarizes the works of this dissertation and points out the direction of the further study.