Research On Random Fault Detection For Networked Control Systems

Networked control systems(NCSs)are closed-loop control systems in which sensors,controllers,actuators and other system components are connected via a communication network.Since the communication resources of actual networked control systems are usually limited,phenomenon such as packet dropouts,delays,nonlinearity and quantization error which are inevitable to degrade the performance and stability of the system or even cause faults will appear.In practical engineering applications,performance and safety of the system are extremely important.If faults cannot be detected in time,it will cause great losses.Therefore,the fault detection problem for NCSs has become the focus of research.The main idea of fault detection is to use the information that can be measured in NCSs and construct a residual evaluation mechanism which is used to detect whether faults have occurred.Under this background,models of systems with random packet dropouts,random time-delays,quantization error and so on in network environment are constructed.Random occurring fault and sensor saturation are also considered,and then the corresponding fault detection problem is studied.The main research contents are as follows:The fault detection problem for a class of networked systems with randomly occurring faults,random packet dropouts and signal quantization is investigated.The measured output of systems is quantized by the logarithmic quantizer.The occurrence of fault and packet dropouts is described by two independent Bernoulli stochastic sequences.Then the fault detection filter is designed.The original fault detection filter design problem is turned into an H_?filtering problem by using state enlargement.Parameters of the fault detection filter are obtained by Lyapunov stability theory and the linear matrix inequality(LMI)method,and the obtained residual signal is used to judge the occurrence of faults.For a class of time-delayed networked systems with random packet dropouts and sensor saturation,random fault detection problem is investigated.Considering the occurrence of fault,packet dropouts and delays meets the Bernoulli distribution,a fault detection filter design method is proposed under the condition that the measurement output is subject to sensor saturation.Firstly,the model of the filtering error system is constructed.Then the sufficient conditions which make the filtering error system exponentially stable in the mean square and achieve the H_?performance are derived on the basis of Lyapunov stability theory and LMI method.Finally,the parameters of fault detection filter are obtained by the convex optimization algorithm.The random fault detection problem for a class of nonlinear networked control systems with random packet dropouts and delays is studied.A unified model which simultaneously describes packet dropouts and network-induced transfer delays has been constructed.Sensor saturation is also considered.Sensor saturation and nonlinearity are described by sector-nonlinearities,and then the fault detection filter is designed.By constructing the Lyapunov function,the parameter matrices of fault detection filter are obtained by LMI method and optimal algorithm,so that the filtering error system satisfies the given H_?performance index.Finally,based on the obtained residual signal and the established residual evaluation mechanism,it can be determined whether the fault has occurred.