摘要
重新讨论了Hopfield神经网络平衡点的全局指数稳定性 ,得到了四个新判据 ,且列举了相应的例子来说明 .由这些定理和例题的证明方法 ,阐述了Lyapunov函数的选择 ,网络参数电容Ci、电阻Ri 及神经元ui(x)的导数u′i(x)上下确界Mi,mi,i=1,2… ,n的处理及不等式放缩的精度在构造网络全局指数稳定性判据中的重要性 。
In this paper,we carefully discuss the conclusions about exponential stability of Hopfield neural networks of the past.Four stability criteria are afresh structured and explained by examples.By the proof methods of these theorems and examples,the author discusses the importance about the choice of Lyapunov function,the treatment of the net parameters as electric capatity,electric resistance,electric voltage and inferior and superior of the output functions' derivatives,and the precision of inequality.The estimate of Lyapunov exponent on the the convergence rate about the global exponential stability is given out.
出处
《许昌学院学报》
CAS
2003年第2期1-7,共7页
Journal of Xuchang University
基金
国家自然科学基金(No .6 9874 0 16 )
高校博士点基金(No .970 4 872 2 )资助课题
