By using semi-discretization technique, a discrete analogue of stochastic Cohen-Grossberg neural networks is formulated. Firstly, the existence of pth mean almost periodic sequence solutions and pth moment global exponential stability of the above semi-discrete stochastic system are investigated with the help of Krasnoselskii’s fixed point theorem and some analysis techniques in stochastic theory. Secondly, the exponential stability and stochastic stabilization for a special discrete stochastic Cohen-Grossberg neural networks are studied. These findings show that stochastic disturbances and small discrete step length have negative effects on the existence of mean almost periodic solutions and moment exponential stability. But they have positive effects on the exponential stability and exponential non-stability for some special models, respectively. Besides, this paper also finds that some unstable neural networks should become exponentially stable by stochastic disturbances. In the end, some examples and computer simulations are given to demonstrate the effectiveness of the theoretical results.

通过使用半离散化技术，制定了随机的Cohen-Grossberg神经网络的离散类似物。首先，p th的存在意味着几乎周期序列解和p借助Krasnoselskii不动点定理和随机理论中的一些分析技术，研究了上述半离散随机系统的瞬时矩全​​局指数稳定性。其次，研究了特殊离散随机Cohen-Grossberg神经网络的指数稳定性和随机稳定性。这些发现表明，随机扰动和较小的离散步长对均值几乎周期解和矩指数稳定性的存在具有负面影响。但是它们分别对某些特殊模型的指数稳定性和指数不稳定具有积极影响。此外，本文还发现，一些不稳定的神经网络应受到随机干扰的影响而变得指数稳定。到底，