Research On Fault Diagnosis And Fault Tolerant Control Via Intermediate Estimator

With the rapid development of the industry,the industrial system is becoming more and more large-scale and complicated.In practical engineering,the faults may bring huge financial loss and social loss to the social society.As a consequence,the higher demand on the safety and reliability of the system has motivated the need for fault diagnosis and fault tolerant control(FTC).As an important discipline of fault diagnosis technique,the fault estimation(FE)techniques have attracted a lot of interests from the control community due to its capability of acquiring the exact information of the faults such as size and shape.On the other hand,since a better control performance can be obtained by compensating the effect of the faults online,the FE based FTC has a significant meaning in practical engineering.In recent years,a number of results can be found on the research of the aforementioned two types of problems,however,there are still many problems need to be resolved.On the basis of the previous work,this dissertation focuses on the main challeng-ing problems in the field of FE and FTC,as well as in the area of robust fault estima-tion,robust tracking control,and fault tolerant control of network systems.The main contents are outlined as follows:(1)The FE problem for a class of Lipschitzian nonlinear systems with single faults is investigated,a novel intermediate estimator is proposed,then an augmented intermediate estimator is constructed to deal with the case with multiple faults.The proposed method has overcome the restriction of the SPR condition,which is required to be satisfied by the conventional adaptive observer.(2)The FE problem for a class of interconnected dynamical systems is studied,based on the relative output measurements from neighbors,for each subsystem,a group of dis-tributed intermediate estimator is proposed to estimate the state and the faults.By ap-plying spectral decomposition technique,sufficient conditions that guarantee the global estimation error system to be stable are derived.(3)The fault tolerant tracking control problem for a class of MASs with multiple faults and mismatched disturbances is investigated.The general case where the leader's input is nonzero and not available to any follower is considered.The mismatched dis-turbances are separated into matched part and unmatched part,the effect of the matched disturbance is compensated by the estimates,while the effect of the unmatched distur-bance is mitigated by adjusting some specified parameters,therefore,better robustness can be ensured.(4)The robust tracking control problem for a class of MASs under a directed graph is studied,a distributed intermediate estimator based tracking protocol is established,where a special form of the observer gain and a decentralized type of controller based on esti-mates are designed,which overcomes the design difficulties aroused by asymmetry of the Laplacian matrix.Sufficient conditions for the tracking error to be uniformly ultimately bounded are derived.(5)The observer based FTC problem for a class of linear uncertain system with sensor measurement errors is investigated,an intermediate estimator based FTC method is provided,where the robustness can be ensured by adjusting some specified parameters,without introducing any performance index.(6)The FTC problem for linear systems with multiple faults and disturbances is studied.For the case with matched disturbance,an intermediate estimator parameter ad-justment FTC method is proposed,where the estimation performance and the control-performance can be ensured by fully exploiting the design freedom of some specified pa-rameters.For the case with mismatched disturbances,an intermediate estimator based H? FTC method is proposed,to obtain better robustness,the mismatched disturbances are divided into matched part and unmatched part,the effect of matched part is compen-sated by the derived estimates,while the effect of the unmatched part is attenuated by introducing an H? performance index.Finally,the results of the dissertation are summarized and further research topics are pointed out.