State Reconstruction And Sliding Mode Control For Stochastic Dynamical Systems With Limited Communication Capacity

Control systems design with limited communication capacity is one of the most important research issues in international control theory domain. However, the proposed approaches and results in the existing work are mostly focused on linear system and linear control methods, and there is little attention on state estimation and stabilization, where external disturbance, model uncertainty and unknown nonlinearity are taken into account simultaneously. It is well known that sliding mode control is insenstive completely to the matched external disturbance, model uncertainty and nonlinearity, and the design and analysis of state estimation and siding mode control with limited communication links has an important and wide applications in practical engineering. However, network communication delay, data packet losses and signal quantization cause difficulty and challenge to traditional H∞ filtering and sliding mode control theories. Combining with sliding mode control theory and digital control theory, this thesis analyzes and investigates the problem of network-based filtering and sliding mode control for complicated nonlinear dynamic systems. The detail content of this thesis is as follows:The H∞ filtering problem is studied for a class of nonlinear stochastic systems subject to sensor saturation over unreliable communication channel. The investigated plant is described by a class of stochastic systems with global Lipschitz nonlinearities and random noise depending on state and external-disturbance. The characteristic of sensor saturation is handled by a decomposition approach. The communication links between the plant and filter are unreliable network channels, and the effects of output logarithmic quantization and data packet losses are considered together. A full-order filter is designed by employing the incomplete output measurements, such that the dynamics of the estimation error is guaranteed to be stochastically stable. Both filter analysis and synthesis problems are investigated, and the explicit expression of the desired filters is also provided.The problem of robust fault detection is investigated for Markovian jump linear systems with infinite distributed delays and unreliable communication links. The effects of signal quantization and measurement missing are taken into consideration simultaneously. A stochastic variable satisfying the Bernoulli random binary distribution is utilized to model the phenomenon of the measurements missing. The aim is to design a fault detection filter such that, for all unknown input and incomplete measurements, the error between the residual and weighted faults is as small as possible. A sufficient condition for the existence of the desired fault detection filter is established in terms of a set of linear matrix inequalities.The filtering problem is addressed for a class of stochastic jump systems with statedependent noise. The network communication links between the plant and filter are impact, and the effects of network-induced transmission delay and sensor saturation and are taken into simultaneous consideration. The main difficulty in this filtering problem is that there exists transmission delay in the received mode signals of the filter side, which results in that the real-time information of jump mode rkis not accessible. To overcome this obstacle, an augmentation approach is developed, based on which the resulting filtering dynamics is modelled as a new Markovian jump system with two jumping parameters. A mode-dependent filtering scheme is then developed to guarantee that the resulting overall system is stochastically stable with a guaranteed H∞ performance index.The robust quantized sliding mode control problem is considered for linear continuoustime systems. The output measurements are supposed to be quantized with a logarithmic strategy before transmitting over the digital channels. The quantized signals(piecewise constants) cannot be used to synthesize a continuous-time sliding mode surface. Taditional observer technique is not effective to handle output disturbances. In this thesis, a filtering-based technique is proposed to solve these difficulties, based on which a slidingmode observer-based control scheme is developed to stabilize the resulting closed-loop systems.The network-based spacecraft attitude regulation control is studied, where both dynamical uniform quantizer and static logarithmic quantizer are employed, respectively.By introducing quantizer parameters in the gain matrix of the sliding mode controller,the quantization effects are compensated by the designed robust quantized sliding mode controller. Under the proposed sliding mode control law, the closed-loop spacecraft attitude control systems is asymptotically stable, and the state trajectory of the spacecraft can arrive on the sliding mode surface in finite time.