Research On Event-triggered Fault-tolerant Control For Cyber-physical System Under Cyber Attacks

Cyber-physical system(CPS)promotes the fusion and interaction between the information dataflow and the physical plant in a higher level,playing a key role in the fields of industry,national defense and the livelihood infrastructure,etc..However,the CPS confronts with the challenges of cyber attacks and the communication resource shortage,the physical layer of the CPS exists issues like the system nonlinearity,stochastic disturbances and the uncertainties,which influence the stability of the CPS.In view of these points,this dissertation investigates the event-triggered fault tolerant control from the aspects of the system control theory,which is expected to improve the network resource saving and the system security.The main research work and contributions are given as follows:1.The dynamic event-triggered fault tolerant control problem is investigated for the cyber-physical system under the random deception attacks.Firstly,an output based dynamic event-triggered mechanism is proposed,in which the error weight matrix is introduced,and it will be co-designed with the controller.Furthermore,the dynamic output feedback based predictive fault-tolerant controller is formulated,which reconstructs the system states,and minimizes the influence of the attack even under the worst cases.Then,the sufficient conditions that guarantee the algorithm recursive feasibility and the constraints of the control input are presented and proved,also it is proved that the closed-loop system satisfies the H₂/H_?performance index in mean square.2.Considering the cyber-physical system with dynamic attacks,a dynamic memory triggered adaptive fault-tolerant control strategy is proposed.Firstly,a dynamic memory event-triggered mechanism is presented,in which the memory term is introduced to reduce the conservation of the triggering condition.In the meantime,the completeness and the feasibility conditions of it are demonstrated and proved.Besides,an information cached based dynamic attack model that charaxterizes multiple attack patterns is proposed,whose aggressivity and the stealthy properity are analyzed quantitatively.Furthermore,a dynamic memory triggered adaptive fault-tolerant controller with the switched adaptive compensation law is constructed,which promotes the suppress the damaging effects better.Based on the Razumikhin lemma and the Lyapunov stability analysis,the closed-loop system is proved to be input-to-state practically stable.3.For the stochastic nonlinear cyber-physical system under Denial of service(Do S)attacks,the dynamic memory triggering fault-tolerant control issue is studied.First,a dynamic memory event-triggered mechanism with the memory regulation factor is proposed,and the completeness,superiority and the feasibility are analyzed.Considering the coexisting situations of the Do S attack,nonlinear uncertainties and the stochastic disturbaces,the fault-tolerant observer is presented.Furthermore,the triggered fault-tolerant controller is designed,which addresses the intermittent communication interruption influences.From the Lyapunov stability analysis,the closed-loop system is ensured to be semiglobally uniformly ultimately bounded in probability.4.For the nonlinear cyber-physical system hazarded by the actuator attacks,a dynamic event-triggered fast finite-time fault-tolerant controller is proposed.Firstly,a dynamic event-triggered mechanism with a fractional power term is formulated,which is compatible with the finite-time stability analysis.Besides,the upper bound of the triggering error tolerance is enlarged.Furthermore,by introducing fractional power terms in the adaptive regulation law,the fast finite-time fault-tolerant controller is formulated,and the attack suppression performance is improved.According to the finite-time stability analysis,it is proved that all the states of the closed-loop system will enter into the neighborhood of equilibrium within finite time.5.For the interconnected cyber-physical system tampered by the actuator attacks,a decentralized dynamic triggering fast finite-time fault-tolerant control approach is proposed.Firstly,a decentralized dynamic event-triggered mechanism is designed,which removes the requirments of the input-to-state stable assumption,besides the Zeno-behavior won't occur.Furthermore,the adaptive dynamic surface based fault-tolerant controller is presented,in which the fractional power terms are added in the control input and the adaptive law,and the response rate against the attacks in the subsystems is enhanced.Based on the finite-time stability analysis,the closed-loop system is guaranteed to be practically fast finite-time stable.