Research On Security Control For Event-Triggered Neural Networks Subjected To Cyber-Attacks

Neural networks are a kind of complex large-scale dynamic systems,which are considered as one of the effective tools for effectively approximating nonlinear systems.Thus,it can be widely applied in many fields such as image processing,pattern recognition,artificial intelligence,and signal processing,etc.Nowadays,the research and application of neural networks has been a research hotspot in networked control systems.The limitation of network bandwidth can bring some problems to networked control systems such as data packet loss,network-induced delay and data security and so on.These problems may lead to the network burden even collapse,and affect the process of normal information transmission.In order to save the limited network resources occupied by the transmission of unnecessary data,scholars introduce the event-triggered scheme to networked control systems.The research on control synthesis problem has received great attention for neural networks with event-triggered scheme.As the introduction of communication network to neural networks,it can improve the timeliness and convenience of data transmission.However,due to the openness of the communication network channel,the networked control systems are vulnerable to cyber-attacks,which can affect the stability and performance of the systems such as reducing the stable convergence speed.The research on the security control for delayed neural networks is not rich enough.Therefore,the research on stability and applications of delayed neural networks has important theoretical and practical significance.By taking network resource constraint and cyber-attacks into consideration,the main work of this paper is analyzing the stability of the delayed neural networks.Then the design of controller and state estimator is studied.The main research contents are given as follows:The first chapter introduces the research status of neural network control system is introduced.The research status of data transmission scheme and security problems is introduced for networked control systems with resource constraint.Besides,the research of control synthesis problem is introduced for delayed neural networks.The required lemmas and symbolic explanation are given.In chapter 2,an unified complex cyber-attacks model is constructed,which contains three common cyber-attacks including deception attacks,replay attacks and Do S attacks.Then,the design of adaptive event-triggered controller is studied for neural networks subjected to complex cyber-attacks.This paper employs the adaptive event-triggered scheme and quantization mechanism to solve the resource constraints.Next,by considering effects of Do S attacks,the entire interval is divided into several intervals,a reasonable Lyapunov-Krasovskii functional is constructed on different intervals.The stability condition of system is acquired by utilizing Schur completement and Jesen inequality.Then the design of controller can be obtained.Finally,the usefulness of the controller can be demonstrated by Simulink of Matlab.In chapter 3,the estimation fusion problem is addressed for event-triggered neural networks with complex cyber-attacks.The complex cyber-attacks model is also adopted,and an event-triggered scheme is employed to reduce the burden of network.By taking measurement noise,external disturbance,time delays and cyber-attacks into consideration,a state error system is established.By constructing a reasonable Lyapunov function,the sufficient conditions of stability and performance are acquired for the system.Then,the design method of estimator gain and fusion parameter is obtained.At last,a numerical example of neural networks is utilized to verify the validity of the results.The chapter 4 summarizes the main research content of this paper.Meanwhile,the shortcomings and future work of this paper is given.