Research On Fault Detection And Fault Tolerant Control Of Networked Control Systems

Networked control systems(NCSs)are feedback control ones which consist of the real-time communication networks.There are some problems in the network communication,such as signal to noise ratio(SNR)constraint,quantization,network delay,bandwidth limita-tion and data dropout,and these problems will affect the system performance and even cause instability.Fault diagnosis and fault tolerant control,as the important technologies which guarantee the reliability of systems,are the effective guarantee of the systems'safety oper-ations.Therefore,it is of great significance to investigate fault diagnosis and fault tolerant control of NCSs with different network communication problems.By combining descriptor system theory,dynamic output feedback control,robust filter,and so on,this thesis is con-cerned with the problems of fault detection and fault tolerant control for NCSs where SNR constraint and quantization are considered.The main content can be summarized as follows:(1)The problems of fault detection and fault tolerant control for linear discrete-time NCSs with SNR constrained channels are investigated.First,a filter-based fault detection scheme is proposed by using a descriptor system approach.In this scheme,an augmented fault detection filer is designed such that the residual generation system is admissible,and the estimation of the fault is obtained.Simultaneously,a dynamic threshold is constructed based on input,external disturbance and stochastic characteristics of the channel noise,a constant?is introduced to reduce the conservatism of threshold,and fault alarm rate of the threshold is also given.Subsequently,on the basis of fault detection,in order to guarantee the post-fault system's robust stability,a dynamic output feedback fault tolerant control strategy is established based on the fault estimation information.The fault tolerant controller and fault estimation are designed separately,which is convenient to solve parameters.Finally,a networked two-tank system is provided to verify the feasibility of the proposed method.(2)For the uncertain discrete-time NCSs with modeling errors,considering the system outputs are transmitted through a SNR constrained channel,a robust fault detection method based on H_?filtering is proposed.First,for the uncertain discrete-time systems with mod-eling errors and SNR constraints,an optimal residual generator with no modeling errors is designed as the reference model,and the robust fault detection filter design is further for-mulated as an H_?filtering problem.Then,a norm-based residual evaluation function is proposed,and an on-line dynamic threshold computing method is also given.Simulation results show the effectiveness of the proposed approach.(3)For the discrete-time Lipschitz nonlinear NCSs with additive Gaussian white noise channels subject to SNR constraints,the fault detection problem is investigated.A residual generator based on the mixed H_-/H_?performance index is designed to generate the residual signal,and the sensitivity index H_-is used to represent the minimum effect of faults on the residual signal,while the robustness index H_?is used to express the maximum influence of unknown disturbance and channel noise on the residual signal.Furthermore,based on the fault detection logical relationship,a threshold on the basis of the upper bound on the modulus of the solution of Lipschitz nonlinear systems and the stochastic properties of channel noise is designed.Finally,a motor system is given to illustrate the applicability of the proposed method.(4)The problem of fault detection for a class of linear discrete-time quantized feedback systems is studied,and the system outputs are quantized by a uniform quantizer and a non-uniform one,respectively.A novel approach is proposed to deal with the quantization error,i.e.,the quantization error is considered as the unknown input of the residual generator.Then,an observer-based optimal residual generator is designed to guarantee a suitable trade-off between the sensitivity to faults and the robustness against disturbance as well as quantization error.Furthermore,two different thresholds based on the stochastic properties of quantization errors are designed for these two quantizers.