利用状态空间分解方法,探讨一类具有特殊激励函数的高阶Cohen-Grossberg神经网络的多周期性问题.该类神经网络的激励函数包括带有饱和区的非递减函数以及一般的细胞神经网络激励函数等.给出了保证此类网络的周期环在饱和区内局部指数收敛的充分条件.所得结果表明,一个n维网络可以有2n个局部指数收敛的周期环存在于饱和区.最后以一个数值例子说明了所得结果的有效性. The state space decomposition method is used to investigate the multi-period problem of a class of high-order Cohen-Grossberg neural networks with special excitation functions.The excitation functions of such neural networks include the nondecreasing function with saturation region and general cellular neural network excitation Functions, etc. The sufficient conditions for the local exponential convergence of the periodic rings of such networks in the saturation region are given. The results show that there are 2n periodic cycles in which the local exponential converges in an n-dimensional network in the saturated region. Finally, Numerical examples illustrate the validity of the results obtained.