Research On Fault Diagnosis For Lipschitz Nonlinear Systems

With the rapid development of the modern industrial technology,the scale and com-plexity of engineering control systems are increasing.However,in practical engineering systems,actuator,sensor,or component faults frequently occur.The occurrence of any kind of faults may bring about performance degradation or even some unpredictable loss-es.Fault diagnosis technique is an important approach to improve the safety and reliability for dynamic systems.Considering the fact that practical engineering systems are almost nonlinear systems,it is very meaningful to study the fault diagnosis problem of nonlin-ear systems.In recent years,the research on this problem has attracted a widespread attention,however,there are still many problems to be resolved.For example,for the Lipschitz nonlinear systems,there is no valid fault detection method considering the fre-quency range of the fault,etc.On the basis of the previous work,this dissertation focuses on the problem of fault diagnosis for Lipschitz nonlinear systems,where different kinds of faults,performance indices,observer/filter design are considered.Furthermore,the results are generalized to the sector-bounded nonlinear systems.The finite-frequency performance indices are in-troduced in terms of the input/output signals and the corresponding fault detection method based on these definitions is given for Lipschitz nonlinear systems.Then,the proposed method is extended to the nonlinear networked control systems.The problem of event-triggered fault detection filter design for discrete-time Lipschitz nonlinear networked sys-tems with finite-frequency specifications is investigated.Based on the idea of state aug-mentation,the unknown input observer and proportional-integral observer are designed respectively for the Lipschitz nonlinear systems with single fault and multiple faults.The problem of fault estimation in finite-frequency domain is studied.For the sector-bounded nonlinear systems with sensor stuck fault,a multi-mode based fault detection method,which is dependent on the servo inputs,is proposed.The theoretical proofs are given for the main results.Moreover,simulation experiments are carried out on some practical system models,such as an inverted pendulum system,a robot manipulator model and a F404 aircraft engine model.The simulation results demonstrate the effectiveness of the proposed methods.This dissertation is divided into eight chapters.The main contents of each chapter are given as follows:Chapters 1-2 summarize the development and main research methods on fault diag-nosis and Lipschitz nonlinear systems.Preliminaries about the considered problem are also given.Chapter 3 investigates the problem of fault detection for Lipschitz nonlinear sys-tems in finite-frequency domain.Based on the use of a reformulated Lipschitz property,the nonlinear error dynamics are transformed into a linear parameter varying(LPV)sys-tem.Two new definitions of finite-frequency H? and H_indices are introduced from a time-domain perspective and sufficient conditions are derived to guarantee that the cor-responding system can achieve such a performance.Subsequently,the proposed fault detection observer design is formulated as a multiobjective optimization problem with linear matrix inequality constraints.Simulation results show the effcctiveness and merits of the proposed method.Based on the results in Chapter 3,Chapter 4 investigates the problem of event-triggered fault detection(FD)filter design for discrete-time Lipschitz nonlinear net-worked systems.The event-triggered transmission scheme is introduced to mitigate the utility of limited network bandwidth.The fault detection filter with the finite frequen-cy H_/H? performance is designed.By introducing slack variable techniques,sufficient conditions for the design of FD filter are derived in terms of linear matrix inequalities.The proposed design method can significantly reduce the data transmission and achieve a bet-ter FD performance than the existing methods.Simulation results show the effectiveness and merits of the proposed method.Chapter 5 investigates the problem of sensor fault estimation for continuous-time Lipschitz nonlinear systems with finite-frequency specifications.First,the sensor fault is considered as an auxiliary state vector and a descriptor system is established by using the method of state augmentation.Based on the reformulated Lipschitz performance and finite-frequency H? performance,a novel nonlinear unknown input observer is designed to simultaneously estimate the system states and the fault signals.Then,the observer design conditions with less conservativeness are presented in the form of linear matrix inequalities.Compared with the existing results,no limitations or previous information of the considered sensor fault is required.The proposed method can estimate the sensor fault for Lipschitz nonlinear systems with large values of Lipschitz constants.Furthermore,the proposed method reduce design conservatism of the entire-frequency domain.Simulation results show the effectiveness and merits of the proposed method.Based on the results in Chapter 5,Chapter 6 investigates the problem of fault estima-tion for discrete-time Lipschitz nonlinear systems with multiple faults.A multi-objective unknown input proportional-integral observer is designed,which can not only estimate the system states and sensor fault,but also automatically adjust the actuator fault online.The proposed method makes full use of the frequency information of disturbances and faults to reduce design conservatism of the entire-frequency domain.Simulation results show the effectiveness and merits of the proposed method.Chapter 7 investigates the fault detection problem for a class of sector-bounded non-linear systems with sensor stuck fault.Different from the existing results,a novel multi-model based fault detection scheme which is dependent on the servo inputs is proposed.Only the relationship between the residuals and the servo inputs is considered.The gener-ated residuals are sensitive to servo inputs in faulty cases and robust against them in fault-free case.Thus,arbitrary small sensor stuck faults including outage cases can be detected by the proposed method.Further,in order to improve the fault detection performance,the sensitivity and attenuation of the servo inputs are characterized in the finite-frequency domain.Simulation results show the effectiveness and merits of the proposed method.Finally,the results of the dissertation are summarized and further research topics are-pointed out.