Research On Data-Driven Control Under Network Attacks

With the integration of network and information technology development,network control systems(NCSs)have been successfully applied to industrial fields.Network control systems have the advantages of simple installation,low operation and maintenance cost,less wiring and it is widely used in smart power grid,aerospace,automatic vehicles and other industrial fields.The introduction of network makes the system easy to realize remote control and resource sharing.However,some malicious network attacks often occur in the process of network transmission.Serious network attacks may cause system paralysis.On the other hand,with the continuous expansion of the industrial scale,the process has become complex,and it is increasingly hard to build a precise mathematical model of the system,and how to design the controller for the network control system under the condition of unknown model is also an important problem.Therefore,this paper combines model-free adaptive control(MFAC)and iterative learning control(ILC)methods to study the security control problem of NCSs under data-driven framework.The main research contents are as follows:(1)For a class of unknown dynamic nonlinear systems,the MFAC problem under periodic denial of service(DoS)attack and random packet loss is studied.Firstly,the model of DoS attack is built.Secondly,the MFAC control algorithm is designed based on the DoS attacks model.The stability of the system in the sense of mathematical expectation is proved through derivation and theoretical analysis.In addition,a compensation control algorithm is designed to offset the negative impact of network attacks on the system.Finally,numerical simulation and inverter system simulation are used to verify the effectiveness of the algorithm.(2)The event-triggered MFAC problem for a class of nonlinear system under intermittent DoS attack is studied.To economize the limited bandwidth resources,an event-triggered mechanism is designed between the sensor and controller.Firstly,intermittent denial of service attack is modeled,which is described as limited by frequency and duration.Then with the help of the dynamic linearization method,an event-triggered MFAC algorithm is designed.Subsequently,the stability of the system is proved by mathematical derivation and theoretical analysis.Finally,the numerical simulation and wheeled mobile robot simulation verify the validity of the algorithm.(3)For a class of repetitive nonlinear systems,ILC under hybrid network attacks is studied.Firstly,false data injection attacks and intermittent DoS attacks are modeled.Combining with the hybrid attack model,the ILC algorithm under hybrid network attack is designed considering the characteristics of repetitive system.Secondly,the convergence of the ILC algorithm is analyzed by mathematical derivation and theory.Finally,the numerical simulation and agricultural vehicle simulation verify the validity of the algorithm.