Schedule And Control Co-design For Networked Control Systems

With the rapid development of computer, network and communication technology, the architecture of control systems has been changing. Networked Control Systems (NCSs) incarnates enough the development tendency of control systems, i.e. network, integration, decentralization, intelligence. As a consequence, NCSs are widely applied. However, the inclusion of network makes the analysis and design of control systems more challenging. Although NCSs already have a lot of research achievements so far, these studies generally consider controller design and scheduling separately, which can not be both organic integration of research and therefore can not guarantee a good performance of the overall system. NCSs scheduling and control Co-design is conducive to the overall system performance optimization. Because it's collaborative network operating performance, bandwidth restrictions, and other network factors into control system design and performance optimization, at the same time, the control system design benefits for the network realization. Therefore, this dissertation will aim at three kinds of NCSs, i.e. model-based NCS, uncertain-NCS and MIMO-NCS. With more attention is paid to the bandwidth constraint of NCSs, the system scheduling and control Co- design is investigated in this dissertation.This dissertation contains the following main sections:1. For linear time-invariant plant, considering a kind of model-based NCS with bandwidth constraint and control performance, the method of the system scheduling and control Co-design is investigated. First, at the smart sensor side, we design a new stochastic communication logic scheduling strategy based on Poisson Process with time-dependent intensity. Under this strategy, the system only needs finite time state update; hence the quantity of transmission message is reduced. With the proof that the stochastic communication logic is essentially a Markov chain, the NCS is modeled as a jump system and the sufficient condition of stability for state feedback system is presented as well. Under the given stochastic communication logic, based on the update time, the controller is given in term of a LMI .The simulation result shows that the scheduling strategy can decrease the network traffic, while the controller can guarantee certain good system performance.2. For the NCS with linear time-varying uncertain controlled plant and bandwidth constraint and control performance. We investigate the method of the system scheduling and control Co-design. An event driven Send on Delta (SOD) scheduling strategy is innovativly used in this kind of NCS, which transmits data based on the importance of the system output. Under this way, the unimportant information will not be transmitted; hence the sensor data traffic is reduced. Regarding the effect of the scheduling strategy as the uncertain of the system output, the system model is obtained. In order to get the estimated state of the system, a guaranteed cost observer is implemented. Under the SOD scheduling strategy, the sufficient conditions of a guaranteed cost controller is given in terms of LMI. The guaranteed cost controller makes not only the NCS stable but also the performance index less than an upper bound. This Co-design method will reduce the system bandwidth constraint as well as enhance the system robustness. Simulation results show the good performance of the proposed method.3. For linear time-invariant controlled plant, a kind of MIMO NCS with bandwidth constraint and control performance is considered. We investigate for the method of the system scheduling and control Co-design. The communication sequences which make the input and output available in a time-sharing mode is proposed to reduce the effect of bandwidth constraint, and therefore the system model is presented. Considering the possible change brought by the communication sequences to the system architecture, the system reachability and observability condition is analyzed and the method used to select the communication sequences which can guaranteed the reachability and the observability is proposed. Furthermore, the sufficient condition of stability for output feedback system is presented. Finally, under the periodic communication sequences, the NCS is transferred to a periodic system. Then a LQG controller is designed. The sufficient conditions for the existence of an optimal controller are obtained in terms of dynamic programming solution. Simulation result shows that using the proposed method the bandwidths constrain of MIMO-NCS is solved and the controller can guarantee good performance of the overall system.