Research On Actuator Fault Estimation For Nonlinear Control Systems

With the increasing complexity of modern control systems,there is an increasing demand for the security and reliability of the dynamic system.It is as a result that fault diagnosis and fault tolerant control for control systems have been receiving more and more attention and extensive research.It is known that fault diagnosis is the foundation of fault tolerant control.How to detect the faults in the system,fast and accurately,is the main task of fault diagnosis.Only in this way can we take appropriate methods to deal with the faults effectively.Fault diagnosis mainly consists of fault detection,fault separation and fault estimation.In the field of fault diagnosis,fault estimation has stirred renewed research interest in recent years.It combines functions of fault detection and fault isolation effectively.In practical engineering,as a result of unexpected model un-certainties,time delays,disturbances,perturbations and noises may occurring in the fault systems,it is quite difficult to obtain the accurate size of the fault from an FDI scheme only.Fortunately,fault estimation can depict the size and shape of the fault and can thus automatically perform the required fault detection.Fault estimation is mainly to design a fault estimator to depict the fault signal so as to realize fault detection and isolation,and the information on the fault can be available.However,in the research field of fault diagnosis,few research papers about fault estimation can be seen.The main reason lies in the difficulty of the quantitative analysis of the fault and a small amount of available technology.In recent years,although some research results on fault estimation have been published,there are still more problems in this field that are needed to be studied in depth,such as the designing of the fault estimator and the robustness in the fault estimation sys-tem.On the other hand,according to the components of the system,the faults in the control system can be roughly divided into actuator fault,sensor fault,controlled object fault and controller fault.In the control system,actuators and sensors are the weakest link and the fault often occurs on these devices.In particular,the actuator failure will generally affect the control strategy of the entire system.Therefore,there is a higher demand for the reliability of the actuator and the research of actuator fault diagnosis in control system is of great practical significance.The actuator fault estimation has stirred renewed research interest in the field of fault diagnosis.In general,actuator faults can be divided into effectiveness actuator fault and additive actuator fault.As a result of that model uncertainty,time delay and unknown disturbance may exist in complex systems,a good fault estimator must only be sensitive to faults and not affected by these factors,so that the fault can be obtained quickly and accurately.This thesis focuses on the problems of actuator fault estimation for three different types of nonlinear control system:nonlinear systems via T-S fuzzy models,nonlinear systems based on linear parameter varying models and Lipschitz nonlinear systems.Five different fault estimators for nonlinear systems in consideration are proposed in this thesis.The main work and research results are as follows:1.The problem of fault estimation and dynamic output feedback fault tolerant con-trol for a class of T-S fuzzy control systems with time-varying delay and actuator faults is considered.A k-step fault estimation method for systems in consideration is proposed,by which,the obtained estimates can better depict the size and shape of actuator faults.Then,based on the on-line estimates,a dynamic output feedback fault tolerant controller is designed to compensate the fault effect on the fuzzy delay system,so that the system is asymptotically stable(in disturbance free,constant fault)while satisfies the prescribed H? performance.Finally,the simulation results show the effectiveness and merits of the proposed method.2.The problem of fault estimation for a class of T-S fuzzy control systems with constant time delay and actuator faults is considered.A fault estimation method via the minimum norm least squares solutions(MNLSS)is proposed,by which,the inverse ef-fect from the derivatives of measured outputs is able to be adjusted effectively in error systems to optimize the fault estimation observer,so as to improve the estimation accu-racy.Based on the proposed MNLSS-based fault estimation method,the stability condi-tion without equality constraints for the error dynamics can be easily converted into linear matrix inequalities(LMIs),which are less conservative than the existing ones.The de-signed estimation observer gains guarantee the asymptotic stability(in disturbance free,constant fault)of the error dynamics with time delay while satisfies the prescribed H?erformance.Finally,the simulation results illustrate the effectiveness and merits of the proposed method.3.The problem of fault estimation for a class of Takagi-Sugeno(T-S)fuzzy sys-tems with local nonlinear parts and actuator faults is concerned.The system matrices in consideration contain unknown components,whereas in most of the literature,the system matrices are required to be known completely.An input-output based fault estimation(IOFE)approach for the fuzzy system is proposed.This method does not depend on the LMI condition,but is automatically adjusted by the input and output information so as to obtain the estimated value of the fault.The estimation error for time-varying faults can asymptotically converge to zero in the absence of disturbances.The estimation conver-gence is proved theoretically.Numerical examples are given to show the effectiveness and merits of the proposed method.4.The problem of fault estimation for a class of linear parameter varying(LPV)Markovian jump systems(MJSs)with actuator and sensor faults is considered.Using the measured output and its adjusted derivatives,an observer based fault estimation scheme is presented for the LPV MJS.A min-max regulator(MMR)based on the Cholesky de-composition technique is designed for the first time to adjust the parameters in the fault estimator,so that the states,actuator and sensor faults are able to be estimated simulta-neously with good accuracy.Then,the designed observer gains guarantee the stochastic stability of the overall error system with a prescribed H? performance.Finally,a nu-merical example is given to illustrate the effectiveness and advantages of the proposed MMR-based fault estimation methods.5.The problem of fault estimation for a class of nonlinear systems with Lipschitzian nonlinearities and faults is studied.A fault estimation method via a full-column-rank state variable substitution approach is proposed,which is able to deal with the fault estimation for that the coefficient matrix of the fault vector is non-full column rank.The designed estimation observer guarantees the asymptotic stability of the overall error system with a prescribed H? performance.The states and faults can be estimated with good accu-racy simultaneously.The proposed estimation approach is further extended to solve the problem of fault estimation for nonlinear systems with multiple faults or/and subject to disturbances.Finally,numerical examples illustrate the effectiveness and advantages of the proposed method.