Stability criteria for Memristor-based generalized neural networks with time-varying delay via an improved integral inequality

This paper focuses on the problem of stability analysis for Memristor-based generalized neural networks (MGNNs) with time-varying delay. The conventional memristor-based neural networks (MNNs) can be regarded as a special case of MGNNs. The paper proposes an improved auxiliary function-based integral inequality to address the derivative of the triple integral within the Lyapunov-Krasovskii functional (LKF). This advanced inequality encompasses well-known integral inequalities as special cases and reduces conservatism by incorporating additional free matrices. Based on the improved integral inequality, the stability criteria with less conservatism for MGNNs are derived by making full use of the information of state and activation function. Finally, the feasibility and superiority of the proposed methods can be demonstrated with three numerical examples.