连续型联想细胞神经网络局部指数稳定性研究

由于细胞神经网络的潜在应用前景 ,它现已成为神经网络研究的新热点。首先给出连续型联想细胞神经网络的数学模型 ,得到了连续型细胞神经网络平衡点局部指数稳定的充要条件及平衡点指数吸引域的估计 ,研究表明对平衡点的指数吸引域的估计 ,只要计算平衡点处的导算子矩阵的对数范数即可。该研究对连续型联想细胞神经网络的设计和应用均有重要的作用。

The research of cellular neural networks (CNN) has become a hot topic in the area of neural networks for its great potential applications. This paper presents the mathematical model of continuous associate memory cellular neural networks (CNN). The sufficient and necessary condition of the local exponential stability and estimation of attractive domain of equilibrium point of this model are obtained. The research indicates that the logarithmic norm of the derivative matrix in the equilibrium point is only needed for the estimation of attractive domain of equilibrium point of this model. The research will have an important significance for the design and application of the continuous associate memory CNN.

国家自然科学基金 (60 0 72 0 3 4) ;;江苏省青年科技基金 (BQ2 0 0 0 0 0 3 ) ;;江苏省高校自然科学研究计划项目 (0 1KJB5 2 0 0 2 );;中国科学院沈阳自动化研究所机器人学研究室基金 (RL2 0 0 10 8)资助课题