Further Improvements on Stability Analysis for Neural Networks with Traditional Delay and Leakage Delay

Weiru Guo,Fang Liu,Runmin Zou
DOI: https://doi.org/10.1109/cac59555.2023.10450679
2023-01-01
Abstract:In neural networks (NNs), there exist two kinds of time delay, leakage delay (LD) and traditional delay (TD). Up to now, much effort has been done to the effect of TD on the stability of NNs while the effect of LD is always ignored. This paper aims at the stability problem for NNs considering both LD and TD by Lyapunov-Krasovskii approach. First, the model of NNs considering both TD and LD is established. A novel Lyapunov- Krasovskii functional (LKF) containing the information among TD, LD and system states is constructed. Then, the novel improved reciprocally convex inequality (NIRCI) is developed. It is employed to transform the stability conditions into the form of linear matrix inequalities (LMIs). The proposed stability criteria can be utilized to check whether the neural network system is stable conveniently. Finally, two numerical examples are done to demonstrate the advantages of proposed criteria and the effectiveness is proved by the simulation experiment.
What problem does this paper attempt to address?