Research On Nonfragile Filtering And Control Of Complex Systems

Among the existing research on filtering and control of complex systems,it is usually implicitly assumed that the filter/controller is ideal,that is,it is designed based on accurate signals and can be perfectly implemented:for one thing,the gain perturbation of the filter/controller in the actual implementation process is not considered,for another thing,the quantization error in the transmission process of various signals is ignored.On the basis of previous studies,this dissertation focuses on the subject of nonfragile filtering and control for complex systems.By fully considering the parametric uncertainty of complex systems,the gain perturbation in the process of filter/controller implementation,external disturbance,potential failure,and the dynamic quantization phenomenon in the process of signal transmission,this dissertation has systematically studied the problems of resilient filtering,nonfragile quantized output feedback control,nonfragile quantized fault detection filtering and fault-tolerant control for complex systems,and obtained the following original research results:1)The problem of resilient H_∞filtering for linear systems with norm bounded uncertainties is studied.Different from some existing results,this dissertation assumes that all filter gain matrices have additive perturbation,and then the decoupling problem of multiple uncertain product terms between model parameters and filter gains is systematically investigated through three improved approaches,namely linear fractional transformation representation approach,norm scalar bound a pproach and slack matrix approach,and the sufficient design conditions of resilient filter are given.2)The problem of nonfragile peak-to-peak filtering for uncertain fuzzy systems is studied.With the help of fuzzy modeling technology,the state estimation problem of a class of parameter controlled nonlinear tunnel diode circuit systems is transformed into the resilient peak-to-peak filtering problem of fuzzy systems.Then a new peak-to-peak performance analysis criterion is given by introducing additional variables.Finally,the problem of multiple uncertain product terms is successfully solved by using the uncertainties separation two-step approach.This approach can also be extended to the design of resilient H_∞filter for fuzzy systems to reduce its design complexity and remove the positive semidefinite constraints in some existing results.3)The problem of nonfragile fault detection filtering for uncertain linear systems under the influence of dynamic quantization is studied.This is the first time to consider the design problem of resilient quantized fault detection filter.Firstly,the residual signal is generated through a fault detection filter based on the quantized output signal,in that way,the considered problem is described as the performance optimization problem of a augmented filtering error system.Then,the nonlinear coupling problem between quantization error and various uncertainties is solved by uncertainties separation two-step approach and S-procedure.Finally,through one-step design method,the corresponding synchronous design conditions of fault detection filter and dynamic quantizer are given.4)The problem of nonfragile static output feedback control for linear systems under the influence of dynamic quantization is studied.Firstly,the more practical case is considered where the input and output are synchronously quantized and the controller gain is equipped with some perturbation due to imperfect implementation.Then,Chang-Yang decoupling approach is used to solve the coupling nonlinear problem in the design process of static output feedback control,S-procedure is used to deal with the quantization error,and a new dynamic quantization design strategy based on equality adjustment rules is proposed.Finally,the synchronous design conditions of output feedback controller and dynamic quantizer are given by one-step design method.The output feedback control problem with two-channel signal quantizations considered in this dissertation is more in line with practical application,and the proposed method for quantization error is simple and universal.5)The problem of nonfragile fault-tolerant control for linear systems under the influence of dynamic quantization is studied.Based on the redundancy of dynami c quantization input signal,a new quantized closed-loop control system is constructed by descriptor representation approach,and the control problem of suspension system is transformed into the multi-objective optimization problem of closed-loop system.Then,a simple and unified quantized feedback control design strategy is given through the Projection lemma,which can be directly extended from state feedback to static output feedback and avoid some equality constraints in some existing results,in addition,a new dynamic quantization design strategy based on interval adjustment rules and a simple adjustment parameter transmission protocol are given.Finally,the synchronous design conditions of controller and quantizer are given.