Observer-based Fault Estimation And Accommodation For Dynamic Systems

Due to the increasing security and reliability demand of actual industrial process control systems, thestudy on fault diagnosis and fault tolerant control of dynamic systems has received considerableattention. Fault accommodation is one of effective methods that can be used to enhance systemstability and reliability, so it has been widely and in-depth studied and become a hot topic in recentyears. Fault detection is used to monitor whether a fault occurs, which is the first step in faultaccommodation. On the basis of fault detection, fault estimation is utilized to determine online themagnitude of the fault, which is a very important step because the additional controller is designedusing the fault estimate. Compared with fault detection, the design difficulties of fault estimationwould increase a lot, so the study on fault estimation and accommodation is very challenging.Although there has been some study results on fault estimation and accommodation for dynamicsystems, the common methods at the present stage have design difficulties, which limit applications ofrespective design approaches. Therefore, the problems of fault estimation and accommodation areneeded to be in-depth studied. This dissertation studies observer-based fault estimation andaccommodation for dynamic systems, and establishes a systemic and comprehensive framework offault estimation and accommodation for continuous/discrete-time systems. Main contributions of thisdissertation are as follows:Firstly, for the conventional adaptive fault estimation (CAFE) algorithm, the performance of faultestimation can not be met, a novel fast adaptive fault estimation (FAFE) algorithm is proposed, whichevidently enhances the fault performance, including rapidity and accuracy. Then, the proof process isimproved to eliminate the strict equation constraint by using inequality transform. Furthermore, theproposed FAFE algorithm is extended to time-delay systems. Based on the slack-matrix design idea,fault estimation of fast time-varying time-delay systems is studied to deal with the difficulty that theCAFE algorithm can only be used in slow time-varying time-delay systems, and fault estimation ofneutral delay systems is addressed to treat the difficulty of the CAFE algorithm can not be used insuch systems.Secondly, for a class of specific faults of loss of actuator effectiveness, a fast fault estimationalgorithm for such kind of faults is proposed. Meanwhile, based on the on-line obtained fault estimate,a fault accommodation scheme is proposed. Note that, the presented fault estimation observer andfault accommodation are designed separately such that the design process can be greatly simplified.When faults of loss of actuator effectiveness occur, the provided fault accommodation scheme canguarantee the asymptotic stability of the whole system.Thirdly, for adaptive observer design, the error dynamics is needed to satisfy the strictly positive real (SPR) condition, a novel full-order fault estimation observer (FFEO) design is proposed, and adynamic output feedback-based fault accommodation is provided. The FFEO possesses widerapplication scopes compared with adaptive observer. Then, a dynamical output feedback-based faultaccommodation design is proposed to avoid design difficulties caused by observer-based statefeedback fault tolerant control. Meanwhile, the results on continuous-time systems are extended todiscrete-time systems, and it is obtained that the corresponding results of discrete-time systems.Fourthly, on the basis of FFEO design, a reduced-order fault estimation observer (RFEO) approach isstudied, and a static output feedback-based fault accommodation is provided. The RFEO possesses thewider application scope of the FFEO, and the on-line fault estimator generated by the RFEO containsthe current output information to enhance the performance of fault estimation. Then, based on theslack-matrix method, a static output feedback-based fault accommodation is proposed. Furthermore,the obtained research results of continuous-time systems are extended to discrete-time systems, and theresults of fault estimation and accommodation for corresponding discrete-time systems are derived.Fifthly, for the research status that the study of fault estimation and accommodation forTakagi-Sugeon (T-S) fuzzy models based nonlinear systems estimation is very few, the fault estimationand accommodation approaches that are proposed in above chapters are extended to T-S fuzzy modelsbased nonlinear systems, the design of fault estimation and accommodation for T-S fuzzy models areobtained. Our studies involve continuous and discrete-time systems, and enrich the content of this studyfield.Finally, in order to verify the practical value of the proposed theoretical methods, a part of theproposed theories is applied to the three degrees of freedom helicopter flight control platform. Thegiven fault estimation design can real-time online estimate actuator faults, and based on the online faultestimate, the designed fault accommodation restores the system performances. Experimental resultsshow that the obtained research results to improve the flight control system reliability and security havesome theoretical reference value, and have broad application prospects.