Event Triggered State Estimation Of Complex Neural Networks With Time Delay

Complex valued neural networks(CVNNs)are an extended form of real valued neural networks.Its design is based on complex valued variables and complex valued algorithms,which can be used to process more complex information.In recent years,it has been widely used in wind prediction,complex phenomenon classification,color face recognition,and other fields.It is worth noting that in these applications,due to the periodic transmission of signals,resources are wasted,and the event trigger mechanism(ETM)effectively solves this problem.Based on this,this thesis mainly studies the state estimation of complex neural networks by introducing new static event trigger control mechanism and dynamic event trigger mechanism for several types of time delays.The main research contents include:1.The state estimation problem of complex neural networks with mixed delay under event triggering mechanism is studied.Firstly,an event trigger mechanism is designed based on the measured output to effectively reduce the update frequency of the estimator.The waiting time is introduced into the trigger mechanism to avoid the Zeno phenomenon in sampling.By using the Lyapunov method and the properties of the complex value matrix,a sufficient criterion for the global asymptotic stability of the estimation error system is established.Based on the linear matrix inequality(LMIs)technique,the algorithm for solving the complex value gain matrix K is given.Finally,a numerical example and its simulations are presented to illustrate the effectiveness of proposed approach.2.The problem of state estimation for a class of discrete-time complex valued neural networks with both leakage delay and discrete-time time-varying delay is studied.The signal transmission from the output sensor to the state estimator is realized through a shared wireless network with limited communication resources.In order to reduce the consumption of communication resources in the communication process,a dynamic event trigger mechanism is introduced to determine when the output measurement should be updated.By constructing an appropriate LyapunovKrasovskii function,a sufficient criterion to ensure the asymptotic stability of the estimation error system is derived,in which the complex neural network is not divided into real and imaginary parts.Using the feasible solutions of a group of linear matrix inequalities with complex variables,the gain matrix of the estimator is designed.A numerical example and its simulation results are given to illustrate the validity of the theoretical result.3.The problem of μ state estimation of mixed delay discrete complex neural networks with event triggering mechanism is studied.The mixed time delay includes infinite distributed time delay and bounded time-varying time delay.The event trigger mechanism is used to determine whether the output signal is updated and transmitted to reduce signal transmission.By using the Lyapunov method and the properties of the complex value matrix,a sufficient criterion for the global μstability of the estimation error system is established.Based on the linear matrix inequality technique,an algorithm for solving the complex value gain matrix K is given.Finally,a numerical example verifies the correctness and effectiveness of the theoretical results...