Adaptive Control Of Strict-Feedback Uncertain Nonlinear System Under Cyber Attacks

Cyber-Physical Systems(CPS)are the next generation intelligent systems which integrate computation,communication and control.In recent years,CPS have attracted the attention of the research community due to the widespread applications.However,CPS are vulnerable to malicious cyber attacks since more open and shared communication networks are used.Attackers can interfere with communication,destroy data or obtain privacy,which results in the influence of physical system,performance degradation,service interruption,or even system failure.Although there have been a lot of achievements in CPS secure control,the proposed control method could not be applied to nonlinear systems(especially strict-feedback uncertain nonlinear systems)since the control objects are mainly linear systems in existing works.Adaptive control strategy has become a very popular control method for uncertain nonlinear systems,which can greatly reduce the difficulty of analysis and control about complex systems.Therefore,it is of great significance to utilize adaptive control to design effective control strategies for nonlinear systems to mitigate or counteract the interference of malicious attacks and prevent serious performance degradation and loss.Based on adaptive control,this thesis studies the problems of two typical attacks,deception attacks and denial-of-service(Do S)attacks,which can inject false data into the system and prevent the transmission of measurement or control signals.The main contents and contributions are shown as follows:(1)The cooperative consensus control of a group of high-order nonlinear multiagent systems under deception attacks is considered.Firstly,the mathematical model of the attack occupying in controller-to-actuator communication link and sensor-tocontroller communication link is established,hence the dynamic equation of the multiagent system under attack is obtained.Then combining the backstepping method and tuning function,a fully distributed adaptive controller is designed under the condition that only the compromised data damaged by the attacks can be obtained.The strategy overcomes the difficulty of obtaining accurate system state information from sensors and solves the problem that the existing method proposed for linear system or relatively simple nonlinear system cannot be applied to multi-agent system with uncertainty.It is shown that all closed-loop signals are uniformly bounded and consensus errors are also bounded in presence of deception attacks.(2)The problem of consensus tracking control for strict-feedback uncertain nonlinear multi-agent systems subject to disparate deception attacks is studied.By defining the new auxiliary variables,the problem that attacks between multi-agents no longer have simultaneity and consistency is solved.Then the variables containing consensus and estimation errors is established,which is instrumental in stability analysis.The virtual controllers are designed according to backstepping technology.A fully distributed adaptive control method is designed,which do not rely on the original state information but only uses the disturbed state signal by attackers.The results indicate that the proposed strategy can resist negative influence derived from deception attacks.It is shown that all the signals in the closed-loop multiagent system are globally uniformly bounded and consensus errors converge to a arbitrarily small compact set in presence of more general cyber attacks.(3)The output feedback control problem of strict-feedback nonlinear system under deception attack against actuator and sensor is investigated.By assuming that only the contaminated system output is available for feedback control,a set of K-filters is applied to develop a novel observer to reveal the unavailable state information.Then,adaptive stabilization control algorithm is proposed based on backstepping method.The introduced filter variables compensate the disturbance triggered by attacks in controllerto-actuator communication link,which plays a core role in restoring system performance.At the same time,only the output of the system subject to attack is utilized for feedback control.In this way,recursive estimation of different parameters due to deception attacks in each step of backstepping technique is avoided,which dramatically reduces calculative resources.It is shown that system output converges to a ball with a radius arbitrarily small and global boundedness of all the closed-loop signals can be guaranteed using backstepping method even in presence of adversarial attacks.(4)Adaptive control for a class of uncertain nonlinear systems under Denial-ofService attacks is investigated.The influence of DoS attacks on system performance under three scenarios is considered.We analyze the close-loop system stability under Do S attacks in terms of attack duration,attack frequency and post-attack duration respectively.Firstly,it is shown that if the duration of each attack meets the corresponding condition,asymptotical convergence of system output can be guaranteed.Secondly,under the scenario that the bounds on the frequency and duration of each attack with respect to overall intervals meet certain conditions,the proposed event-triggered control scheme guarantees that all the closed-loop signals are globally bounded and the stabilization error converges to a ball with a radius arbitrarily small.Thirdly,if the post-attack duration is greater than a given constant after an arbitrarily long attack,closed-loop boundedness is still preserved.The theoretical analysis of the impact on system stability under three scenarios from Do S attacks is established.We overcome the difficulty that the discontinuous sampling state from sensor makes the virtual controllers non-differentiable.Meanwhile,the coupling problem between closed-loop stability analysis and parameter estimation subject to Do S attacks is solved.Compared with the existing work,a more comprehensive stability analysis is obtained.In conclusion,the proposed adaptive control algorithms under cyber attacks solve the secure control problems.Brief summary and future research works are given at the end of the thesis.