Analysis And Design For Networked Control Systems Based On Time-varying Sampling Periods

Networked control systems (NCSs) are a type of closed-loop systems, where the communication networks are employed to transmit information and control signals among control system components (sensors, controllers, actuators, etc.). NCSs are the development and integration of control science, computer technology and communication technology. Compared with the traditional control systems, NCSs have many advantages such as sharing resources, reduced weight, lower cost, better interactivity, less cost of installation, easier maintenance, higher efficiency, flexibility and reliability, etc. However, the introduction of the communication networks leads to network-induced delays, packet-dropouts and time-varying sampling period inevitably, so the use of network may deteriorate the performance or cause instability, which makes the analysis and synthesis of NCSs complicated.In this dissertation, the modeling, stability and robust control problems of time-varying sampling period NCSs with network-induced delays and packet-dropouts are deeply studied by applying the robust control and stochastic control theory, which are mainly investigated by employing theoretical analysis and numerical simulation methods. The main results obtained in this dissertation are summarized as follows:(1) Modeling and control of NCSs with network-induced delays and packet-dropouts are investigated which are based on active time-varying sampling period. Suppose the sensor is both time driven and event driven, the controller and the actuator are event driven, the sampling periods switch arbitrarily in a finite set, the number of packet-dropouts is random and bounded. The linear time-invariant system through network is modeled as a discrete time-varying delay system by iterative method and active time-varying sampling period method. Based on the Lyapunov approach, the sufficient conditions for stochastic stability of NCSs are derived, where the successive packet-dropouts are driven by a finite state Markov chain and an identically independently distributed sequence respectively. The stabilizing controller is constructed in terms of linear matrix inequalities (LMIs) correspondingly. Finally, the simulation example is given to illustrate the validity and feasibility of the results.(2) H? control of NCSs with the partially known packet dropout information is investigated. Suppose the networked-induced delays are random and bounded, the network exists sensor-to-controller as well as controller-to-actuator. Since the successive packet-dropouts are driven by an independent finite Markov chain, the NCSs with the designed state feedback controller are modeled as a Markov jump system. Aming at the fact that the packet dropout information is hard to be obtained accurately, namely it does not need to know the transition probabilities completely, the sufficient condition for stochastic stability with H? norm bound yhkf is presented based on Lyapunov method. The results of the numerical examples demonstrate the H? performance better over the existing methods.(3) The asymptotic stability and strictly dissipative control problems for NCSs with packet-dropouts are studied which are based on passive time-varying sampling period. Suppose the sensor, the controller and the actuator are all time driven, the packet-dropouts have both an upper bound and a lower bound, the sampling period is time-varying and fluctuates across the nominal period. The NCSs are modeled as a class of discrete-time system with parametrical uncertainties by the parameter uncertainty method. The asymptotic stability conditions are given according to the Lyapunov method. Based on it, an improved Lyapunov-Krasovskii function is constructed and some new sufficient conditions for strict (Q,S,R)-dissipativity are derived via the LMIs formulation by using the discrete Jensen inequality. The numerical examples have shown that the designed method is less conservative and reduced computational complexity.(4) The stabilization problem is investigated for a class of NCSs with random delays and random sampling periods. Sampling periods can switch randomly in three cases according to the high, low and medium three types of network load. The sensor-to-controller and controller-to-actutor random delays and random sampling periods are modeled by three synchronous and independent Markov chains. The designed state-feedback controller depends on not only the current sensor-to-controller delay ?ksc but also the most recent available sampling period h-(?)sc at the controller node, and the output-feedback controller depends on not only the current sensor-to-controller delay r" but also the most recent available sampling period hk-?ksc and the most recent available controller-to-actutor delay ?k-?ksc-1ca at the controller node, moreover,hk-?ksc depends on ?ksc and hk, ?k-?ksc-1cadepends on ?ksc and ?kca. Based on it, the resulting closed-loop systems are multi-step Markov jump linear systems with multi-mode. The sufficient and necessary condition for stochastic stability of the closed-loop systems is presented, and the controller design is converted into a set of matrix inequalities, which can be transferred into LMIs by using the cone complementarity approach. Based on it, when the transition probabilities are partially known, according to the property of multi-step transition probability, the greater than or equal to three-step transition probability information is not available, and ?s???xr?rs(i)=?s???rs(i)=1(i?3). A new method to deal with unknown elements in multi-step transition matrix is proposed, the sufficient condition for stochastic stability of the closed-loop systems is obtained accordingly. The cart and inverted pendulum example is given to illustrate the effectiveness of the proposed method.