Analysis And Synthesis Of Model-Based Discrete-Time Networked Control Systems

Networked control systems (NCSs) have received much attention recently. An NCS is a control system in which a control loop is closed via a shared communication network. The use of a shared network in the feedback path offers the advantages of low installation cost, reducing system wiring, simple system diagnosis and easy maintenance. However, the NCS also has some inherent shortcomings, such as bandwidth constraints, packet delays and packet dropouts, which will degrade system performance or even cause closed-loop instability. Hence it is necessary to do further researches for NCSs. Based on the previous works on NCSs, this dissertation mainly studies the modeling, analysis and synthesis of NCSs for the packet dropouts and bandwidth constraints.The main contents are outlined as follows.1. The stability of the discrete-time NCSs with polytopic uncertainty are considered where a smart controller is updated with the buffered sensor information at stochastic intervals and the amount of the buffered data received by the controller under the buffer capacity constraint is also random. Sufficient conditions are established for guaranteeing the exponential stability of generic switched NCSs and the exponential mean square stability of Markov-chain driven NCSs, respectively.2. The analysis and synthesis of discrete-time NCSs are studied where the plant has additive uncertainty and the controller is updated with the sensor information at stochastic time intervals. It is shown that the problem is linked to H∞control of linear discrete-time stochastic systems and a new small gain theorem is established. Based on this result, sufficient conditions are given for ensuring mean square stabil-ity of the NCS, and the genetic algorithm is utilised to design the controller of the NCS based on a linear matrix inequality (LMI) technique.3. The separation principles for Markovain jump linear system are derived under single- packet transmission and multiple-packet transmission strategies, respectively. The observer-based discrete-time NCSs are investigated where random packet dropouts occur independently in both the S/C and C/A channels. Firstly, model the C/A and S/C channels as two independent Markov chains and convert the NCSs as Markovain jump linear systems. Secondly, derive the stability condition and the synthesis of stabilisation control by the corresponding separation principle.4. The discrete-time NCSs under the single-packet transmission and multiple-packet transmission strategies are investigated where the plant has time-varying norm-bounded parameter uncertainties, both the S/C and C/A channels experience random packet dropouts. Sufficient conditions for synthesis of robust stochastic stabilisation and design of robust H∞controller are derived in the form of LMIs.The conclusions and perspectives are presented in the end of the dissertation.