State And Fault Estimation Of Discrete-time Linear Parameter-Varying Systems

With the increasing complexity of modern industrial equipments,fault detection and diagnosis(FDD)technology can effectively improve the security and reliability of the system.Currently,the research on state and fault estimation of discrete systems is not enough.Based on the existing studies,this study will focus on the state and fault estimation of discrete-time linear parameter-varying(LPV)systems.The main contributions of this thesis are as follows:An observer is proposed for the simultaneous estimation of the system state as well as actuator and sensor of a discrete-time LPV systems.The presented approach enables disturbance attenuation and guarantees observer convergence.First,by extending system states and sensor fault,the discrete-time LPV systems are transformed into the descriptor LPV systems,it is further transformed into the traditional form of discrete-time LPV systems.Then,an H∞ observer is proposed for the simultaneously estimate of the actuator&sensor faults and states of the discrete-time LPV systems.Then,the problem of observer design is translated into a linear matrix inequality solution.Finally,numerical example and experimental tests on a three-tank system validate the effectiveness of the presented method.A method is presented to simultaneously estimate actuator and sensor faults and states for a class of discrete-time LPV systems with integral measurements and time-delay.First,an augmented state is constructed to combine the current system states,the integral measurements states and sensor fault,the system is converted into a descriptor discrete-time LPV systems.Next,a constructive sufficient condition is presented and proved for the H∞stability of the proposed observer.Then,the simultaneous estimation of the actuator&sensor faults and system state is transformed into the solution of linear matrix inequalities,and the calculation procedure of observer gains is given.Finally,a numerical example validates the effectiveness of the proposed method.