Research On Robust Fault Detection Methods And Application Of Network-Based Switched Time-Delay Systems

The scale of the industrial control system has gradually increased,the complexity has increased day by day in recent years,and once the system fails,it will cause huge economic and material losses.In order to avoid the paralysis of the whole production process caused by failure,more stringent requirements will be put forward for the reliability and safety of the system.In addition,when a large disturbance or fault occurs in switched systems,the behavior of switched systems is more complex due to the interaction between continuous and discrete dynamics in the system.It is necessary to design more effective fault detection and diagnosis technology to avoid the instability and paralysis of switched systems caused by fault.At the same time,with the continuous maturity of artificial intelligence technology,the globalization of industrial robots is intensified increasingly,and the research on industrial robots have been developed rapidly,and its research direction has penetrated into various fields.As one of the important forms of industrial robots,manipulators are widely used in industrial fields such as automobile manufacturing,mold processing,electrical engineering and electronics.If the components of the manipulators are damaged or the sensors and actuators fail to be dealt with in time,it will cause the manipulators to malfunction and result in a decrease in system performance,which puts forward higher requirements for the stability and safety of the manipulators.Meanwhile,the manipulator system is susceptible to external interference during operation,and there are often uncertainties in the system modeling process.At the same time,the manipulator has the characteristics of nonlinearity and strong coupling.Therefore,it is a challenging task for researchers to establish a more accurate mechanical arm system model and design a fault detection strategy with excellent performance.The robust_?performance of network-based switched time-delay systems,the co-design of controller and fault detection filter,and the fault detection method based on intermediate estimator are deeply discussed,and a new set of fault detection methods for network-based switched time-delay systems is formed based on Lyapunov-Krasovskii function(LKF)stability theory and linear matrix inequality(LMI)technology.Some theoretical results are simulated and verified on the actual manipulator system.The conservatism of the obtained results is greatly reduced due to the use of the Jensen inequality method,Wirtinger integral inequality method,improved reciprocal convex combination method,and single summation inequality method based on multivariate auxiliary functions in the processing of time delay.The main work of this dissertation is summarized as follows:(1)The background and significance of the research are introduced,and the development,classification and research status of fault detection and diagnosis technology are comprehensively described.Basic concepts and research background of switched time-delay systems;Research status of fault detection for switched time-delay systems and overview of fault detection for switched manipulator systems.(2)Lagrange method was used to model the dynamics of the switched manipulator system,and a modeling method based on joint torque feedback was used to obtain a more universal dynamics model of the manipulator subsystem,and its properties were analyzed in detail.Then it introduces some basic knowledge and important lemma used in the follow-up research to provide a theoretical basis for the follow-up research.The dynamics of the switching manipulator system is modeled by Lagrange method.At the same time,a more universal dynamics model of the manipulator subsystem is obtained by using a modeling method based on joint torque feedback,and its properties are analyzed in detail.Then it introduces some basic knowledge and important lemmas used in the follow-up research,which provides a theoretical basis for the follow-up research.(3)The problem of fault detection and controller co-design under asynchronous switching is studied for discrete nonlinear switched time-delay systems.Observer design is introduced into fault detection filter(FDF)design based on mode-dependent average dwell time(MDADT)strategy and piecewise LKF method.As a residual generator,discrete nonlinear switched time-delay systems can be transformed into_?the model matching problem,the designed closed-loop FDF can not only make the estimated value between residual and fault as small as possible but also satisfy the exponential stability condition of discrete nonlinear switched time-delay system.A new single summation discrete inequality method based on the multivariate auxiliary function method is adopted to deal with time delay,and the relaxation conditions for the stability of systems are obtained.Then,the event triggered control(ETC)is introduced into discrete nonlinear switched time-delay systems with random packet dropouts,and the cooperative design of FDF and controller is studied.The event trigger is designed,assuming that the process of packet loss satisfies the Bernoulli random distribution sequence,and the asynchronous switching model of FDF and original switching system is established.Based on MDADT method and LKF stability theory,a sufficient condition for exponential stability of nonlinear switched time-delay systems is given.By using cone complementarity linearization algorithm,the FDF parameter solving problem is transformed into a convex optimization problem,which reduces the conservatism of stability results.Finally,simulation results verify the effectiveness of the proposed method.(4)Aiming at the two-DOF manipulator model,the dynamic method is used to model the manipulator as a switched time-delay system.The problem of fault detection and controller co-design for the manipulator switched system with actuator fault is studied.Based on LKF stability theory and ADT switching method,a sufficient condition for exponential stability of nonlinear switched time-delay system is given,and the solution of FDF gain is transformed into a convex optimization problem by using cone complementary linearization algorithm,which reduces the conservatism of stability results.Finally,the simulation results verify the effectiveness of the proposed method.(5)The problem of fault detection and controller co-design for continuous nonlinear switched time-delay systems under asynchronous switching is studied based on the MDADT strategy.Different from the ADT method,the MDADT switching method can make each subsystem have its own residence time,which makes the system more suitable for practical application.In order to improve the degree of freedom of design,a closed-loop fault detection strategy based on the co-design of controller and FDF was proposed.FDF is used as the residual generator through piecewise LKF and MDADT methods,and the continuous nonlinear switched time-delay system can be expressed as a_?model matching problem.At the same time,the derivative of LKF is estimated more accurately using Wirtinger integral inequality and an improved reciprocal convex combination method,the relaxed stability result and a less conservative result are obtained,and the problem of solving FDF and controller parameters is transformed into an LMI feasibility problem.Finally,the proposed method is applied to an actual manipulator test platform to verify the results.The effectiveness and superiority of the proposed method are verified by the simulation examples of high maneuverability aircraft technology and the experimental results of manipulator.(6)A novel method based on an intermediate estimator is proposed to study its fault estimation problem aiming at linear networked control systems with time-varying delay and packet loss.The proposed method can not only eliminate the observer matching constraint but also estimate the state and fault.Then,LKF stability theory and LMI technique are used to solve the gain of fault estimation filter and obtain sufficient conditions for exponential stability of dynamic error system.Based on the above conclusions,the theoretical results are extended to the fault estimation of nonlinear networked switched system with time-varying delay and packet loss.Assuming that the process of packet loss satisfies the Bernoulli random distribution sequence,a sufficient condition for exponential stability of switched time-delay system is given and proved based on the average dwell time(ADT)method and LKF stability theory.At the same time,the congruence transformation method is used to remove the design constraints to obtain less conservative feasible results.Finally,the proposed method is applied to an actual manipulator test platform to verify the results.Numerical simulation and manipulator experiment results demonstrate the effectiveness of the proposed method.Finally,the research work of the whole dissertation is summarized,which shows the existing problems and future development direction in the research of fault detection of network-based switched systems.