Research On Event-triggered Fault Diagnosis Of Discrete-Time Stochastic Systems

Motivated by the development of sensors,wireless devices,microprocessor chips,dynamic stochastic systems in wireless network environment have become a research target of industry and academia.However,the constraints of node power supply and complex network environment restrict its application,which prompts this paper to research on fault diagnosis in non-ideal network environment.Therefore,event-triggered wireless network transmission mechanism has received extensive attention.In the non-ideal network environment,many problems need to be solved in the wireless network with event-triggered transmission strategy.In this paper,we focus on the event-triggered fault diagnosis of discrete stochastic systems,the following contents are studied.(1)Fault detection of stochastic network systems is studied in this section.The module of residual generation and evaluation for fault detection is described,and two energy norms are proposed to make fault detection residual have the best sensitivity to fault and robustness to interference injected by attackers and false information.In addition,the stochastic Lyapunov function is used to solve the weighted matrix of estimator gain and residual.The performance of fault detection is validated by application examples and simulation experiments.(2)Fault estimation for nonlinear stochastic systems based on event-triggered mechanism is studied.The estimator and event-triggered mechanism are designed for the nonlinear stochastic system with state and sensor saturations.Lyapunov function and linear matrix inequalities are used to obtain the gains of fault estimator and the eventtriggered condition.In addition,the strong nonlinearity in the process of sensor modeling is approximated by the radial basis function neural network method,and the above theory is simulated,which shows the validity and potential application value of the theory.(3)The problem of fault diagnosis for discrete Markovian jump systems is studied.On the one hand,a new event-triggered fault estimation method and corresponding eventtriggered conditions are proposed to reconstruct system state and actuator fault under event-triggered data transmission.On the other hand,an event-triggered fault-tolerant controller based on state estimation and actuator fault estimation is proposed to ensure the closed-loop performance of the system.With the help of linear matrix inequalities,the gain of the controller is derived by stochastic stability theorem,and the effectiveness of the method is illustrated by simulation examples.