Robust Filtering And Control Of Nonlinear Switched Systems With Non-gaussian Noises

Switched systems belong to an important category of hybrid systems,which are composed of a set of subsystems?continuous or discrete?with a switching signal that schedules between them.This type of system provides a more flexible frame for modeling complex systems.At the present stage,a lot of results have been achieved for the research on linear switched systems.However,due to the existence of more or less nonlinear and time-varying delay characteristics in actual systems,the stability and controllable of the switched systems have a serious impact.Because the Takagi-Sugeno?T-S?fuzzy model has unique advantages in describing the smooth nonlinear of the systems,and Gaussian noise belongs to a kind of ideal external noise of the systems.Therefore,the robust control and filtering for switched systems affected by the Non-Gaussian with mainly studies in this paper,and based on this research related issues such as the design of robust controllers and filters with different performance indicators are discussed.It is also considered that the dynamic characteristics of actual systems are difficult to accurately describe as linear.Because linear parameter varying?LPV?systems greatly reduce the difficulty of tuning parameters compared to other gain scheduling techniques,in this thesis was further based on LPV switched systems done some research.The main innovations and research results of this thesis are shown as follows:First of all,for a nonlinear time-delay switched systems described based on T-S fuzzy model,when the controller parameters are perturbed to design the non-fragile robust H_?state feedback controller and the robust H_?filtering when the system states immeasurable are studied.The principle of Parallel Distributed Compensation?PDC?is used to obtain a closed-loop T-S fuzzy switched systems,and the Average Dwell Time?ADT?switching strategy is utilized to design the switching signal of the systems.Lyapunov-Krasovskii functional stability theory and Moon inequality technology are used to deal with the time-delay dependence in the systems,and the design problems of controller and filter are converted into convex optimization problems of LMIs that are easy to be solved by computers,which further guarantees the closed-loop systems and augmentation systems are asymptotically stable.The effectiveness of the method is verified by simulation examples.Secondly,take into full consideration that the external input and output of the designed systems are energy-peak stability,for the time-delay T-S fuzzy switched nonlinear systems,the input noise signals are deemed to be energy-bounded and the output error signals are considered to be peak-bounded.The closed-loop systems and filter augmentation systems meet the conditions of L₂-L_?performance constraints,and based on this,a robust L₂-L_?dynamic output feedback controller matrix and L₂-L_?filter matrix solution algorithm are given.The controller and filter parameters are given by LMIs technology,and simulation calculations demonstrate the effectiveness of the design method.Finally,considering the mismatched between the filters and systems in the actual switched systems the variable parameter characteristics phenomena.Thus H_?asynchronous fixed-order filtering problem is studied for the LPV switched systems under the mode-dependent average dwell time?MDADT?switching signal.The stability of the LPV switched systems are studied based on the mode-dependent and parameter-dependent Lyapunov-Krasovskii function,and the projection lemma is used to deal with the parameter couplings in the systems,and then through congruence transformation and variable substitution the filter with H_?performance is obtained.Simulation examples verify that the H_?fixed-order filter designed under the MDADT switching strategy is less conservative than the filter obtained under the ADT switching signal,and the filters obtained by parameter-dependent function are less conservative than the result under the parameter-independent functional.