Passivity Analysis And Control For Cellular Neural Networks With Time Delay

Since cellular neural network has been widely used in various aspects such as pattern recognition,signal processing and optimization problem and so on,the passivity and controller design problem have attracted the attention of many scholars.In the practical application of cellular neural networks,time delay is universal phenomenon,which may adversely affect the performance of system,and may even cause the system to be unstable.Therefore,the passive analysis of cellular neural network with time delay has scientific research value and practical guidance meaning.Based on passivity theory,Lyapunov-Krasovskii functional(L-K functional),integral inequality,reciprocally convex method,linear matrix inequality and other theories and methods,the passivity and feedback controller design of cellular neural network with time delay are studied.The main work is as follows:1.The passivity problem of the cellular neural network system with time delay is studied.First of all,novel and augmented L-K functional is constructed for the system,which not only contains more information about time delay,but also contains two L-K functional terms constructed based on integral inequalities,which relaxes the conditions to ensure that the L-K functional has positive definiteness;Then,two new zero equalities containing double integral terms are constructed,which further expand the application of the zero equation in passivity analysis;Finally,based on above methods,new passivity criteria with less conservativeness are proposed.2.The passivity problem of the uncertain cellular neural network system with time delay is studied.First,based on the existing L-K functional,L-K functional is constructed for the system,which contains more information about time delay and quadratic terms of time delay;Then,zero equalities containing double integral terms are expanded by introducing more information about time delay into the zero equalities;Then,an improved reciprocally convex method and new sufficient condition to ensure that the derivative of L-K functional including quadratic terms of time delay is negative are proposed,which are effective to reduce the conservativeness of the passivity criteria;Finally,more relaxed passivity criteria are given by combining with above methods.3.The controller design problem of cellular neural network system with time delay and packet dropout is studied.First of all,a suitable feedback controller is designed according to the characteristics of the system,which consider more information about the current state and past state of system;Then,based on methods such as time delay division method,L-K functional and linear matrix inequality,the stability of system is studied;Finally,stability criterion with less conservativeness is given.