| Neural networks are a kind of large-scale nonlinear dynamic systems,which are established by simulating the structure of human brain neural networks.Due to the complexity of the actual environment,the structure of neural networks may change with the change of external factors.Therefore,the idea of the switching is introduced to form switched neural network models.Switched neural networks with high interconnection provide a framework for designing large-scale parallel processors,which are widely used in pattern recognition,image processing,associative memory and so on.In addition,the internal state variables that describe the dynamic law of the system often appear unmeasurable in practical engineering.Hence,it is necessary to estimate the internal state of the system according to the available measurement data.In recent years,the state estimation of switched neural networks has attracted more and more attention.This paper mainly investigates the problem of state estimation of delayed switched neural networks.It mainly includes the following contents:Firstly,we introduce the background and status of delayed switched neural networks,the basic structure of this paper,and the partial definitions and lemmas.Secondly,we mainly study the finite-time state estimation for a class of switched neural networks with both leakage delay and time-varying delays.By constructing delay-dependent Lyapunov-Krasovskii(L-K)functional,combining the average dwell time(ADT)condition and the free-weighting matrix method,sufficient criteria for the existence of finite-time state estimator are given.The mode-dependent estimator gains are designed based on linear matrix inequalities(LMIs).A numerical example is given to verify the validity of the results.Finally,the exponential state estimation of switched neural networks under the simultaneous action of leakage delay and time-varying delays is considered.Based on multiply L-K functional method,ADT and free-weighting matrix techniques,delay-dependent criteria for ensuring global exponential stability of error systems are obtained.Both the existence conditions and the explicit characterization of the desired estimator are derived in terms of LMIs.Two numerical examples are given to prove the correctness and validity of the conclusions. |
