Quasi-projective synchronization of inertial complex-valued recurrent neural networks with mixed time-varying delay and mismatched parameters

Abstract

This article investigates the quasi-projective synchronization of inertial complex-valued recurrent neural networks (ICVRNNs) with mixed time-varying delay and mismatched parameters. By using an appropriate variable transformation, the order of the ICVRNN differential system is reduced, and then, by applying the real decomposition method, it is separated into real and imaginary components. The matrix measure approach with the nonlinear Lipschitz activation functions is employed in the ICVRNN model. Through the proper description of the matrix measure approach, some sufficient conditions have been derived for the quasi-projective synchronization criteria of the considered model through designing a suitable controller. Here, some significant results have been provided for the ICVRNNs with mismatched parameters and mixed time-varying delay. Finally, two numerical simulations are discussed to validate the feasibility and persistence of our obtained results with few conditions. © 2022 Elsevier Ltd