Fault Detection For Affine Fuzzy Systems And Fault-tolerant Control For Uncertain Linear Systems

With the development of science and technology,the practical plants become more and more complex,the occurrence of any fault may lead to performance deterioration or even instability of the system,thus causes the incalculable loss.Therefore,it is necessary to improve the reliability and the security of the systems,fault detection and fault tolerant techniques provide the effective ways for this problem.In this thesis,by using the linear matrix inequality(LMI)and adaptive techniques,the problems of fault detection for T-S affine fuzzy systems and fault tolerant control for uncertain linear systems with multiple mismatched uncertainties are investigated,where different kinds of faults and event-triggered mechanism are considered.The main research contents are as follows:For the discrete-time T-S affine fuzzy systems with output saturation,the fault detection observer design problem is studied.On one hand,the adaptive event-triggered mechanism is introduced to reduce the burden of network communication.One the other hand,by considering all possible regions of plant and observer,the problem of mismatched regions which is caused by the immeasurable predefined variables is solved.Based on the piecewise Lyapunov function and the free-weighting matrices method,the design condition of fault detection observer for T-S affine fuzzy systems is given and the effectiveness of the proposed method is validated by the simulation.For the linear systems with multiple mismatched uncertainties,by designing the adaptive fault-tolerant controller directly,the stable operation of systems can be ensured with or without faults.The considered uncertainties include mismatched norm-bounded,affine and polytopic uncertainties.These mismatched uncertainties are handled by utilizing the linear fractional transformation(LFT)and LMI methods,the adaptive technique is used to compensate faults.Based on the cone complementary linearisation algorithm,the controller solving criteria is given.Furthermore,considering the systems are accompanied with stochastic disturbance,utilizing differential mean value theorem and combing with the stochastic Barbalat's lemma,the closed-loop system and error system can be proved to be stable in probability 1.Finally,the simulation of rocket fairing model validates the effectiveness of the proposed method.