Fault Diagnosis And Fault Tolerant Control For Distributed Measurement Systems

Distributed estimation is a fundamental problem in the field of sensor network.Com-pared with traditional centralized estimation,distributed estimation has lower cost and robust performance.Compared with decentralized strategy,it has better performance and wider application range.However,since the local measurement output and the observed system often cannot form an observable system,which brings some difficulties to the design of distributed observer.The distributed fault diagnosis and fault-tolerant control based on distributed observer also faces some challenges.Nowadays,fruitful results of distributed estimation have been presented.However,the existing methods and estima-tor structures are complex.Few research on distributed fault diagnosis and fault-tolerant control has been studied.Motivated by the above discussions,this paper focuses on the research of fault diagnosis and fault-tolerant control based on distributed observer.The main work and research results are as follows:The existence condition and simple design procedure are studied.A minimal order distributed observer for linear systems is also investigated.The proposed distributed ob-server consists of a network of observers.Each observer in the network has access to only part of the output of the observed system,and communicates with its neighbors according to a given network graph.Each local observer can estimates all the states asymptotically.In addition,Each local observer has the same order as the observed system.In order to reduce the order of the distributed observer,we study the existence condition and design procedure of minimal order distributed observer.Some local output measurements are used to construct partial states directly,which reduces the order of the distributed observ-er.Distributed fault detection based on H_-/L_?performance index and interval observ-er techniques is investigated.Firstly,the distributed H_-/L_?fault detection observer is designed to achieve fault detection.H_-/L_?index not only attenuates the effect of distur-bance to the residual,but also increases the fault sensitivity on the residual.Further,L_?generates the threshold for fault detection.In addition,for uncertain systems,this paper proposes a distributed fault detection method based on interval observer.By designing the distributed interval observer,the local output interval of the normal system is estimat-ed.When the fault occurs,the part outputs will exceed the interval and the fault will be detected.The design methods of full-order and reduced-order robust distributed fault estima-tors for linear systems are studied.Firstly,the fault is regarded as the augmented state of the original system.An augmented system is constructed,and then a distributed fault estimator is designed for the augmented system.Each local fault estimator exchanges information with each other through a given communication topology graph.Each esti-mators can simultaneously estimate the states and faults of the original system.In order to reduce the order of the local estimator,this paper further proposes a reduced-order robust distributed fault estimator by using local measurement information.The problem of active fault tolerant control for linear systems based on distribut-ed observers is investigated.Firstly,a distributed active fault-tolerant control method is proposed for additive faults.In this paper,additive faults are regarded as augmented s-tate.Fault is estimated online and then is accommodated by the designed controller based on fault information.The fault-tolerant controller gain and estimator gain matrices are calculated by stabilized the global system.In addition,a distributed active fault-tolerant control method is proposed for the multiplicative fault of the actuator.In this paper,the local information is used to estimate the multiplicative fault of the actuator,and then the fault is compensated locally.The distributed active fault-tolerant control is realized for the global system.