Research On Quantized Feedback Control For Networked Control Systems

With the rapid development of science and technology, Networked Control Systems (NCSs) has been penetrated into many branches of engineering, and the subjects of study are still coming forward. Based on the stability of control network and the minimal performance index of the NCSs, this paper considers the problem of quantization, erasure channel, time-delay, dropout, outside disturbance and the unmodeled dynamics in NCSs. We study the problems of modeling, the design of quantizer, channel model, robust stability, H∞optimal control and guaranteed cost control.First and foremost, a survey is given to illustrate the background and research subjects of NCSs. Based on the comprehensive study, this thesis puts forward the shortcoming of NCSs, and gives the NCSs development tendency. Then, we address the following questions:①We study the problem of modeling for NCSs with quantization, delay, dropout and unmodeled dynamic.②We consider the optimal control in NCSs under the situation where the network channel and quantizers not only exists between the sensor and controller, but also exists between the controller and actuator.③The cooperative design of coder/decoder and channel model in NCSs is studied.By using the Lyapunov function, the problems of modeling and stability analysis are studied in NCSs subject to quantization and delay for continue and discrete system, respectively. Constructing a new dynamic time-variable Lyapunov function that related to the quantization errors and ranges, and the robust stability conditions and the corresponding controller design can be obtained based on LMI.Using Lyapunov-Krasovskii function, for the NCSs with unmodeled dynamics, quantization and delays, we given the delay-independent robust stability conditions, guaranteed controller and H∞optimal control based on LMI.As we know, there are two possible locations in NCSs for quantizers and the channel, one location named measurement path where the quantizer and channel located between sensors and controllers; the other is actuation path where the quantizer and the channel located between controllers and plant or actuator. By using the stochastic Lyapunov-Krasovskii functional and matrix inequalities, a stochastic stability condition and estimator-based controller are provided for NCSs. Another fact is the quantizer can be divided into coder and decoder. Consider the joint effects of coder/decoder and communication channel, we deal with the optimal performance index for two kinds of network architecture, namely, classical control framework and Youla control framework. Based on the Signal-to-Noise ratio (SNR) constrained channel and minimum error variance criterion, the optimal coder/decoder or quantizer is designed include the character of the limited channel.As an application, for automatic control system of autopilot, including the effects of network, we study the problem of rudder roll damp and course keeping controller design with saturated and quantized input.Finally, Conclusion is drawn at the end of this thesis. Further discussion on future work is also made.