Hybrid Triggered Filtering And State Estimation For Delayed Neural Networks With Cyber Attacks

Neural networks are a kind of artificial neural systems which consist of many widely connected neurons.Recently,due to the widely application of neural networks in handling with different kinds of information and distributed information processing,neural networks are widely used in artificial intelligence,deep learning and data mining.At the same time,neural networks have become a hot topic in the area of networked control systems.Considering the existing packet loss,time delay and information security in the environment of traditional networked control system,these problems may lead to the network burden in networked control systems.Therefore,the investigation about stability of neural networks has theoretical significance and application value.In view of network induced delay and information security,this paper concentrates on the investigation about stability of neural networks,the design methods of filter and state estimator are also given.The main contribution of this paper are given as follows:In the first chapter,the research status of networked control systems are given.The data transmission mechanisms and the problem of information security are also introduced in the networked control systems.In the following,the main issues of neural networks which are investigated in this paper are introduced and some lemmas and symbols are also given.In the second chapter,the hybrid triggered scheme is introduced to save limited networked resources for networked control systems which contains the network induced delay and packet loss.The problem of filter design for neural networks is investigated at the same time.On the investigation of information transmission security of networked control systems,the stochastic cyber attacks is considered in non ideal networked environment and the model of delayed neural networks is constructed.By using Lyapunov stability theory and liner matrix inequalities,the sufficient conditions which can ensure the stability of filter error system and the designed filter parameters are provided.Finally,numerical examples are employed to illustrate the design method.In the third chapter,based on the hybrid triggered scheme and quantization,the problem of state estimation for neural networks is investigated with the consideration of stochastic cyber attacks.In order to improve the utilization rate of network resources and save the limited network resources,the hybrid triggered scheme is introduced.Considering the network induced delay and stochastic cyber attacks,the estimating error system is constructed.The sufficient conditions which can ensure the stability of estimating error system and the gain matrix of the state estimator are given by using Lyapunov stability theory and liner matrix inequalities.Finally,matlab simulation example is given to show the feasibility of the proposed method.The fourth chapter summarizes the main work of this paper.The weakness of this paper is pointed out and the research orientation is given in the future.The major innovation of this paper are listed as follows:(1)It is the first time that the hybrid triggered scheme is introduced in neural networks.The hybrid triggered scheme consists of periodic sampling scheme and event triggered scheme.The stochastic switch between these two triggered schemes is described by a variable which satisfying Bernoulli distribution.(2)On the investigation of information security,the stochastic cyber attacks is considered and the sampling data is transmitted safely when the system is stable.