Research On H∞ Quantized Feedback Control For Networked Control System With Markov Characteristics

Networked control system (NCS) is a type of distributed control system where sensors, actuators and controllers are interconnected by real-time communication networks. Due to the primary advantages of NCS such as low cost, simple installation and maintenance, resource sharing and remote control and so on, it has been widely applied in the practical engineering. But, note that some inevitable phenomena when the control signals transmitted through the communication network, several challenging issues will appear such as time-delay, packet dropout, quantization and so on which may inevitably degrade the control performance of NCS, or even cause the system to be unstable. Consequently, the research on NCS with mutilple network induced phenomena is of great theoretical and applicable significance and has appeared to be a topic of significant interest to the control community. There are some random abrupt changes in the NCS due to factors, such as environment changes or disturbances, and Markov chain can well describe the randomness of time-delay and packet dropouts in network control system. With the raise of Markov theory, it has played an important role in solving the stochastic problem. In addition, the control problem of NCS has to be focused on the component fault condition, because the whole system will break down and have serious consequences. The main work of this thesis can be summarized as follows:1. The quantized feedback Hoo control problem is investigated for NCS with time delay and quantization. First, a Markov chain with partially unknown transition probabilities is adopted to describe the random parameters, the NCS is modeled as Markov jump system. The quantizing effects are transformed into sector bounded uncertainties, time-delay is time varying. Then, by constructing a Lyapunov-Krasovskii functional candidate, the sufficient condition based on LMI is given to ensure the system to be stochastically stable and with a H∞ performance level γ, and the H∞ quantized feedback controller is designed by cone Complementarity Linearization (CCL) algorithm. Based on the above model, the uncertain NCS is taken into consideration, the time delay is modeled by Markov chain with partially unknown transition probabilities. The Hoo quantizer feedback controller is designed to make the system stochastically stable and with a Hoo performance level γ. Finally, a numerical example is given to illustrate the effectiveness and efficiency of the proposed design method.2. The Hoo control problem is investigated for NCS with mixed random delays and quantization. Markov delay and distributed delay are both taken into consideration. The NCS is modeled as Markov jump system with partially unknown transition probabilities. The infinity distributed delay is dealed with by constructing a new Lyapunov-Krasovskii functional candidate. The Hoo quantized feedback controller is designed to make the system to be exponential mean-square stable and with a H∞ performance level y. A numerical example is presented to illustrate the effectiveness of the proposed techniques.3. The H∞ filtering control problem is investigated for uncertain NCS. In the study, network-induced delays, quantization and packet dropout are all taken into consideration. The system is modeled as Markov jump system with partially known transition probabilities. The occurrence of each time delay has a certain probability distribution, and the conservativeness of system can be improved by this method. The packet dropouts are described by a Bernoulli process. A full-order Hoo filter and quantized feedback controller are designed to make sure the resulting filtering error system to be mean square stable with a prescribed Hoo performance level base on LMI. A numerical example is given to illustrate the effectiveness and efficiency of the proposed design method.4. The problem of H∞ fault detection is investigated for NCS with quantization and random packet dropouts. In the study, the packet dropouts are modeled by a Markov chain with two models, the system is modeled as Markov jump system. By constructing residual generator, the problem of fault detection is translate into filter design problem. The Hoo fault detection filter is designed to make the resulting residual system to be stochastically stable with a prescribed Hoo performance level γ. A numerical example is given to illustrate the effectiveness and efficiency of the proposed design method.5. The problem of Hoo fault detection and quantized feedback control are investigated for nonlinear NCS with quantization and mixed time delays. First, the Hoo fault detection filter is designed for NCS with Markov delay and distributed delay, the dynamic quantizer is applied to quantize the output signal. Then, base on the above model, the sensor fault matrix is modeled by two state fault model and switched by Markov chain with partially known transition probabilities. The H∞ controller is designed by constructing a Lyapunov-Krasovskii functional candidate to ensure the nonlinear system to be stochastic stable with a prescribed Hoo performance level y under sensor fault. Finally, a mass spring system and remote DC servo motor control system are given to illustrate the effectiveness and efficiency of the proposed design methods.