The Research Of Fault Detection Technology For Networked Control Systems With Time-varying Sampling Periods

Sensors, controllers and actuators are often connected over a network medium, whichare called networked control systems (NCSs). Compared with the traditional controlsystems having point to point structure, advantages of NCSs include flexible structure, highreliability, low cost, easy installation and maintenance, etc. For NCSs, the occurrence ofnetwork-induced delays, data packet dropouts, time-varying sampling periods,communication constraints and other issues are inevitable. Currently, the existing resultsmainly focus on the modeling, stability analysis and controller design for NCSs.Due to the increasing demand for safety and reliability of plant in production practice,fault detection approach has attracted the attention of scholars. Recently, research on faultdetection for NCSs with time-varying sampling periods is still in the start stage. Therefore,this thesis studies the modeling of NCSs with time-varying sampling periods, design of faultdetection filter, design of fault detection system and selection of residual evaluation function.The main contents of this thesis are summarized as follows:(1) Considering packet dropouts between plant and fault detection filter andtime-varying sampling periods, the problem of fault detection filter design for NCSs isstudied. The sampling period switching-based approach and the parameteruncertainty-based approach are adopted to deal with time-varying sampling periods. Thefault detection system is constructed based on the observer-based fault detection filter. Theobserver-based fault detection filter design criteria are proposed; residual evaluationfunction and fault detection logic are given. Finally, the simulation results show that thedesigned observer-based fault detection filter is effective. Even if the controller-to-actuatorpacket dropouts are considered, the proposed fault detection filter design methods are stillapplicable.(2) The problem of fault detection for NCSs with communication constraints andtime-varying sampling periods is studied. By using a Bernoulli distributed white sequence todescribe the influences of communication constraints in sensor-to-controller andcontroller-to-actuator, a new closed-loop model for NCSs is established. Based on theestablished model, design criteria of fault detection filter are proposed; residual evaluationfunction and fault detection logic are given. Suppose that sensor nodes and actuator nodesare allowed to access network to transmit data or not to access network in each sampling period at the same time, the problem of fault detection for NCSs with communicationconstraints is converted into the problem of fault detection filter for NCSs with data packetdropouts in multiple channels. The observer-based fault detection filter design criteria areproposed.(3) The network-induced delays in sensor-to-controller and controller-to-actuatorchannels are considered simultaneously. The system is described by an uncertaindiscrete-time model with time-varying parameters inside a polytope whose vertices areobtained by means of the Cayley-Hamilton theorem. Based on the established model, designmethod of full-order fault detection filter is proposed and sufficient conditions are proposedto asymptotically stabilize the fault detection system.