On Global Asymptotic Stability of Cohen–Grossberg Neural Networks With Variable Delays

This paper studies the problems of existence, uniqueness, global asymptotic stability and global exponential stability of the equilibrium of Cohen-Grossberg neural networks with variable delays. An estimation technique based on delay differential inequality with variable coefficients is developed to establish delay-independent/delay-dependent sufficient conditions for global asymptotic/exponential stability. The stability criteria obtained are based on the M-matrix theory. These criteria can be easily checked in practice and do not require that the delays be constant or differentiable. In particular, our delay-independent asymptotic/exponential stability criteria remove a restriction on the amplification functions imposed by the existing results. Furthermore, our delay-dependent exponential stability criteria give explicitly the allowable upper bound on the diagonal delays such that the global stability property of Cohen-Grossberg neural networks can be retained. Thus, our new results are of great importance in design and application of Cohen-Grossberg neural networks with variable delays. The effectiveness of the new results is further illustrated by two numerical examples in comparison with the existing results.