Exponential Stability of Neural Networks with a Time-Varying Delay Via a Cubic Function Negative-Determination Lemma.
Xu-Kang Chang,Yong He,Zhen-Man Gao
DOI: https://doi.org/10.1016/j.amc.2022.127602
IF: 4.397
2023-01-01
Applied Mathematics and Computation
Abstract:The problem of global exponential stability of neural networks with a time-varying delay is studied in this article. Firstly, to fully utilize the cross-term relationships among state variables, an improved augmented delay-product-type Lyapunov-Krasovskii functional, in-cluding an extra double integral state, is established for the stability analysis. Accordingly, this augmented LKF derivative is a higher-order function of the time-varying delay. Then, three state vectors are considered to reduce the order of the function to cubic. So, to ob-tain the feasible negative-definiteness condition of this LKF derivative of non-convexity, a negative-determination lemma for cubic functions is employed to handle this problem. As a result, a novel stability criterion is obtained. Two well-known numerical examples illus-trate the effectiveness of the criterion. (c) 2022 Elsevier Inc. All rights reserved.