Fault Estimation Observer Design And Fault-tolerant Control For Dynamical Systems

In modern industrial systems,the component faults including sensor and actuator faults may lead to the performance degradation,and even cause catastrophic accidents.Therefore,the safety of the system design and operation plays an important role in practical engineering.On the other hand,with the development of piratical engineering,the structure of system is becoming more and more complex,many systems possess the characterisations of jumping parameters and nonlinearities,which bring some difficulties in the fault-tolerant design.It is of significance to investigate the fault reconstruction and fault-tolerant control problem for the complicated dynamical systems.The detail content of this thesis is as follows:The observer-based fault-tolerant controller is designed for Markovian jump nonlinear system with time-varying delay.An adaptive observer approach is developed to solve the state estimation and FE problem simultaneously.The observer can estimate different kinds of faults accurately,such as constant or time-varying characterization.The derivative of the faults is supposed to be bounded,which means the faults can be arbitrarily large.Based on the estimation,a fault-tolerant control(FTC)technology is proposed to guarantee the stochastic stability of the closed-loop system.As the theorem is given by linear matrix inequality(LMI),it is convenient for us to calculate the designed parameters.The state estimation and FTC problems for a class of semi-Markovian jump systems(S-MJSs)with external disturbance and sensor faults have been studied.The original SMJSs is transformed into two subsystems,where the first one has the external disturbance but free from sensor faults and the second one only includes sensor faults.Then two robust observers are designed for each subsystem.By the discontinuous term in each observer,the effects of disturbance and sensor fault on state estimation are eliminated.The proposed scheme can estimate states accurately without a high dimension in the observer.Based on the estimation,an integral-type sliding mode control(SMC)scheme against faults and disturbances is developed to solve the FTC problem for S-MJSs.It is shown that the proposed FTC method can stabilize the state estimate space and the estimation error space,and the reachability of the resulting sliding mode dynamics is guaranteed.The problem of actuator degradation and additive sensor fault is studied in a unified framework for stochastic MJSs with partially accessible mode information.In the proposed method,the original plant is augmented into a descriptor system,where the state and sensor fault are assembled into the new state vector.A new adaptive robust observer is proposed for the augmented system to estimate the system state,the actuator efficiency factor,and the sensor fault simultaneously.Based on the estimation,a new adaptive SMC scheme against fault and disturbance is developed to stabilize the resulting closed-loop fault system.In the design,the exact knowledge of external disturbance is not required,which is indeed estimated by the online adaptive mechanism.The piecewise fault estimation(FE)and FTC method is developed for discrete-time nonlinear systems under actuator fault,sensor fault,external disturbance and measurement noise.Specifically,the discrete-time nonlinear system is represented by T–S fuzzy models.Firstly,the original fuzzy plant is augmented into a descriptor system,where the state and fault are assembled into a new state vector.Then,a piecewise observer is designed to obtain the accurate estimation of the state and fault.Based on the estimation,a piecewise FTC approach is developed to stabilize the closed-loop system.The time-varying actuator fault problem in network-based nonlinear systems is considered.A general model of unknown time-varying actuator failures is considered in net control systems(NCSs),where the actuator degradation,stuck fault and outage problem are considered in a unified framework.By using the quantized data instead of the original system states,an adaptive fuzzy method is first employed to approximate the unknown nonlinearity.A new adaptive fault-tolerant control scheme is developed to compensate the quantization effects and actuator faults.