Stability And Synchronization Of Switched Memristive Neural Networks

As an important class of hybrid systems,switched systems have received a lot of interest in past decades because of its wild applications,such as the orbit control of spacecraft,traffic control and so on.On another research field,with the successfully made of nanoscale memristor and the development of memristor theory,the memristor-based neural networks become a new research hot topic.Based on the existing results,combining the characters of memristive neural networks and switched systems,this paper analyzed the stability and synchronization control of switched memristive neural networks.The main work is summarized as follows:The input-to-state stability(ISS)problem of switched memristive neural networks with impulsive and random disturbance is studied.In the beginning,we consider two cases,the first is that all subsystem of switched memristive neural networks are unstable,the other is that switched memristive neural networks are consist of both stable subsystem and unstable subsystem.Then,the stability sufficient conditions of switched memristive neural networks in both cases have been given by solving linear matrix inequalities(LMIs),which can be easily solved.The exponential synchronization problem is discussed for a class of switched memristive neural networks with sampled-data controllers.Unlike the general assumption that systems' switching only happened at the sampled instants,we suppose at most one switching occurs during the sampling intervals.Sufficient conditions about the synchronization control problem for the switched neural networks are derived by constructing a switched Lyapunuov function and using LMIs.Moreover,the gain matrices of sampled-data controllers are given.The asymptotic synchronization problem is investigated for a class of switched memristive neural networks.By using the common Lyapunov function and switched adaptive/statefeedback controllers,sufficient conditions that master systems and slaver systems asymptotic synchronized with arbitrary switching are given.