Research On Estimator Design For Networked Control Systems With Packet Loss And Unknown Input

The concept of a networked control system was first introduced by G.C.Walsh and usually refers to a collection of sensors,controllers,actuators and communication networks within a certain area.With the addition of networks,communication between sensors and controllers and between controllers and actuators in control systems often suffers from packet loss or time delay,which can deteriorate system performance and even lead to system instability.Therefore,a growing number of control scholars are committed to designing more optimized estimators to solve the problem of the simultaneous unknown input and state estimation for systems with packet loss or time delay.This thesis will further put forward an effective technical and theoretical analysis scheme,aiming at design a linear minimum mean square error estimator and a state and unknown input simultaneous estimator respectively for network control systems with packet loss and unknown input based on whether the statistical properties of packet loss can be obtained.In addition,for network control systems with time delay and unknown input,the state observer and unknown input estimators are obtained by solving linear matrix inequalities based on state augment method.The main research of this thesis are as follows:(1)For the simultaneous unknown input and state estimation of networked control systems with packet loss,we design two different schemes depending on whether the statistical properties of packet loss can be obtained.The first scheme is based on the statistical characteristics of packet loss,we first model the packet loss process as an independent and identically distributed Bernoulli random process,then,a linear minimum mean square error estimator is obtained by solving a Riccati equation.Another scheme is for the case in which the statistical characteristics of packet loss are difficult to obtain.First,the packet loss is regarded as an unknown input whose any prior conditions are unknown,the transformation of packet loss estimation problems to unknown input estimation problems is further achieved.Then,based on unbiased minimum variance theory,we design the state and unknown input simultaneous estimator in which unknown input estimator is obtained by least squares combined with innovation analysis and state estimator is designed using the Kalman filter estimation principle.Finally,the thesis verifies the designed estimators are very effective in tracking the real state of the system in both cases through MATLAB simulation experiments.(2)For the simultaneous unknown input and state estimation of networked control systems with time delay,we firstly transform the system with time delay into a new delay-free system by using the state expansion method.Then,on the premise that the unknown input and noise disturbance are bounded,the unknown input observer is designed by using linear matrix inequality,and we give the necessary and sufficient conditions for the existence of the unknown input observer.Based on the state estimation,we further design the unknown input estimator and analyze the stability of the dynamic error system.Finally,the effectiveness of the proposed algorithm is verified by simulation experiments.