Secure Estimation For A Class Of Lipschitz Nonlinear Networked Systems Under False Data Injection

With the rapid development of communication technology and computer equipment,networked control systems have come into being.Since the physical devices are connected by network,detecting and controlling systems remotely is realized.Thus,networked control systems are widely used in intelligent medical treatments,power systems,production cooperation networks and other large-scale engineering systems.However,the introduction of the network also brings new challenges,such as packet loss,time delay and quantization error.Moreover,information systems are vulnerable to hackers from inside and outside.Recently,more and more network security incidents of large industrial control system appear.These security events have caused huge economic losses,influence people's daily life,and endanger public security.Therefore,the security of networked control system has been paid great attention.Related achievements are cropping up,which effectively solve a large number of network security problems and ensure the safety and reliability of systems' operation.In this paper,we describe the secure estimation problems in packet-loss environment.The design methods of secure estimation under single attack and multiple attacks are given,respectively.The estimation of false data attack signals and state variables are realized with intermittent observation and control data.The contents include:The first part: we define the problem of attack estimation for a class of Lipschitz nonlinear discrete systems under false data injection attacks.Both the physical system and the packet-loss environment are described in detail.Based on the intermediate variable,the forms of security estimators under the actuator attack and the concurrent attacks(both actuator attack and sensor attack)are given respectively.Then,the corresponding error systems are constructed and the estimation problems are transformed into the stability analysis problems.The second part: this part solves the problem of security estimation under actuator attack.The sufficient condition for the stability of corresponding estimation error system is given through Lyapunov stability analysis.Then,a linear matrix inequality is constructed using Shure complement theorem,and the estimator gain satisfying the design requirement can be obtained.Furthermore,we discuss the selection method of the parameter in the design process of security estimator under actuator attack.Finally,this algorithm is extended to the estimation algorithm under state attack.The third part: this part is dedicated to solve the problem of security estimation under concurrent attacks.The sufficient condition for the stability of corresponding estimation error system is given through Lyapunov stability analysis.Then,the proof process of exponential convergence of state estimation error,actuator attack estimation error and sensor attack estimation error is given.Moreover,the design of the parameter under concurrent attacks is discussed.Finally,this algorithm is extended to the estimation algorithm under direct attacks.The fourth part: we verified the feasibility of the designed methods through the networked motion control system platform.The establishment of dynamic model of the motor is presented,and the data transmission success rate of actuator network and sensor network are obtained through Ping test.Then,the estimation gain is obtained using Matlab toolbox,and the error between estimated value and actual value of actuator attack,sensor attack,motor position,and motor speed are analyzed.Besides,the feasibility and effectiveness of our estimation algorithms are shown by comparison with other algorithms.Finally,the research results of this paper are summarized,and prospective of further research is given.