摘要
通过构造适当的Lyapunov泛函分析了一类时变时滞双向联想记忆神经网络的平衡点稳定性问题。不要求激励函数的单调性和可微性,得到了保证时变时滞双向联想记忆神经网络的平衡点全局鲁棒渐近稳定的两个新判据。所得到的结果能够表示成线性矩阵不等式形式,进而易于用内点算法等方法来验证。通过仿真例子与其他文献中的一些结果进行比较,表明了本文所得结果的有效性。
Based on the construction of Lyapunov function, global stability problems were studied for bi-directional associate memory neural networks with time varying delays. Without monotonicity and differentiation assumptions on activation function, two sufficient conditions were derived for the global robust stability of equilibrium point. The obtained results possess a linear matrix inequality structure and can be efficiently solved by using interior-point algorithms. Comparison between our results and the previous ones through a numerical example was conducted, which implies the effectiveness of the proposed results.
出处
《吉林大学学报(工学版)》
EI
CAS
CSCD
北大核心
2007年第6期1397-1401,共5页
Journal of Jilin University:Engineering and Technology Edition
基金
国家自然科学基金资助项目(60534010
60572070
60521003)
教育部长江学者计划及创新团队资助项目(IRT0421)
关键词
自动控制技术
双向联想记忆神经网络
时变时滞
鲁棒稳定
线性矩阵不等式
automatic control technology
bi-directional associative memory (BAM) neural networks
time varying delays
robust stability
linear matrix inequality (LMI)