Research On Robust Control And Filtering Methods For Nonlinear Networked Control Systems

In recent years,with the continuous development of science and technology and continuous enhancement of network communication capabilities,Because the network control system has the advantages of high flexibility,cost saving and easy resource sharing,it is widely used in the fields of robotics,aerospace and so on.However,the frequent occurrence of data collisions and node failures cause network latency and packet loss during data delivery,which greatly reduces the performance of the system and even destabilizes the system.Therefore,due to the introduction of the network caused by the data transmission delay issues must be considered,and in real life,the delay and packet loss phenomenon is usually random,so the study of the random delay problem has certain theoretical significance and practical value.In addition,it is also one of the most important problems to study the saturation phenomenon existing in the control system and the fault detection of the sensor-saturated system.At the same time,we must also pay attention to the actual project site,the controlled object is often there is some sense of non-linear,using a number of linear models to approximate the original system TS fuzzy model to describe the nonlinear method is not negligible.Due to the continuous presence of continuous peak boundedness of external interference signals in real systems,therefore,this paper studies the nonlinear NCSs from the traditional robust H_? ? L2-L? performance constraints to the more practical robust L₁ performance constraints and validates the effectiveness of the method through simulation examples,the main contents are as follows:Firstly,for non-ideal network environments with stochastic delays,the problem of observer-based robust controller design is investigated.It is assumed that the time-varying delay existing in network environment has a certain stochastic property.By introducing a Bernoulli random variable to represent the distribution probability of delay in different intervals,a new system model with probability distribution information is established.According to the parallel distribution compensation method and Lyapunov stability theory,a robust H_? and L2-L? performance criterion based on T-S fuzzy model is established to make the nonlinear NCSs output be energy bounded and continuous peak bounded.On this basis,the state observer gain matrix and the controller gain matrix are simultaneously obtained by using the linear matrix inequality technique.Finally,the simulation results verify the effectiveness of the proposed method.Secondly,considering the sensor saturation constraints,the robust filter design for a class of nonlinear NCSs with time-varying delay is studied.Assuming that the nonlinear function is affected by the Lipstitz constraint,the parameter uncertainty is bounded by the norm,and the problem of saturation is dealt with using the non-linear fan-shaped region method.The state error augmentation method is used to establish the filtering error system.By selecting theappropriate Lyapunov function,the robust H_?performance criterion and the peak-peak?L₁?performance criterion for nonlinear NCSs under saturated sensor saturation constraints are established.The design of the device transforms into a convex optimization problem for solving a set of linear matrix inequalities.Finally,for a continuous network control system with sensor saturation constraints,the problem of robust fault detection for the system under the condition of bounded amplitude interference is studied.The robust fault detection filter is constructed to generate the residual signal,and the robust fault detection filter design problem is transformed into a robust L₁ filter design problem.According to the Lyapunov stability theory,the delay dependent peak peak?L₁?performance criterion under the constraint of sensor saturation is established.On this basis,the fault signal is detected by the threshold logic method,and the LMI technology is used to solve the design filter.