Event-triggered Based Filtering Of Networked Systems Under Deception Attacks

Networked control systems(NCSs)are a class of closed-loop distributed control systems consisting of sensors,controllers,actuators and control objects connected by communication network,which have the advantages of high noise immunity,low cost and easy maintenance.However,due to the limitation of communication network bandwidth,coupled with the real-time information exchange of multiple objects,there are a series of inevitable problems in NCSs,such as network induced delay,packet dropout,data sequence confusion and so on.Especially in view of the characteristics of sharing,opening and interconnection of communication network,the deception attacks from external malicious attackers seriously weaken the data processing capacity or data storage capacity of NCSs.These phenomena will lead to the deviation of the real-time information of the system,which not only affects the execution of the system control task,but also produces a large number of meaningless signals and consumes the network resources.The introduction of event triggered communication mechanism(ETCM)and signal quantizer into NCSs can obviously reduce the transmission ratio of worthless data,improve the efficiency of signal exchange and optimize the control performance.This thesis mainly analyzes and discusses the influence of various problems in NCSs,such as network induced delay,random measurements missing,external malicious deception attacks,and gives the corresponding solutions.The main contents are as follows:(1)The background and significance of NCSs?ETCM and deception attack are summarized.Aiming at the problems of time-delay,random missing measurements and deception attacks in NCSs,the research results and bottlenecks at home and abroad are introduced.(2)Aiming at the Markovian jump neural networks(MJNNs)under random missing measurements and deception attacks,the stability performance and H_?filtering are addressed.ETCM is proposed to reduce the communication burden of data transmission,and a variable following Bernoulli distribution is used to describe the random phenomenon of missing measurements.Combined with Lyapunov stability analysis and Jenson inequality,the stability criterion of the system is derived and the H_?filter satisfying the conditions is designed.(3)The H_?filter design of discrete time neural networks(NNs)under deception attacks is studied.Aiming at the network burden problem,an adaptive event triggered communication mechanism(AETCM)is proposed,which adjust the trigger threshold according to the dynamic performance feedback to alleviate the communication burden.By constructing a feasible Lyapunov-Krasovskii functional based on the mathematical model,the method of maintaining asymptotic stability of nonlinear systems and H_?filter based on AETCM are provided.(4)The H_?output feedback control of Markovian jump systems(MJSs)under cyber attack is discussed.The time-varying characteristics of the system are simulated by the random jump of Markov chain,and a Bernoulli variable is utilized to describe the randomness of cyber-attack.In order to comb the signals more smoothly and save network resources,the event trigger and quantizer are introduced into transmission channel.Finally,combined with Lyapunov stability theory,the sufficient conditions to ensure the stability of the system and the design parameters of the output feedback controller based on ETCM are obtained.(5)The innovation and shortcomings of the research content are summarized.