机构地区: 华中理工大学控制科学与工程系
出 处: 《数学物理学报(A辑)》 1999年第2期211-218,共8页
摘 要: 研究了Hopfield型随机时滞神经网络dx(t)=[-Ax(t)+Bσ(x(t一τ))]dt+f(t.x(t),X(t—τ))dw(t)的均方指数稳定性与几乎必然指数稳定性.应用Layapunov函数与鞅不等式,建立了这种随机时滞神经网络指数稳定的时滞相关的充分条件.文献中某些关于确定性的时滞神经网络x(t)=-Ax(t)+Bσ(x(t-τ))与神经网络x(t)=-Ax(t)+Bσ(x(t))的稳定准则是文中的特殊情况. In this paper the exponential stability in mean square and almost surely expo- nential stability. are investigated for stochastic neural networks with delay of the form dx(t) = [-Ax(t) + Bσ(x(t - τ) )]dt + f(t,x(t),x(t - τ) )dw(t). For such neural networks, several sufficient conditions for the exponential stability are established by the Lyapunov function method together with martigale inequalities, The obtained results are dependent of the size of delay. Some stability criteria in the literature for the deterministic neural networks with delay x(t) = - Ax(t) + Bσ(x(t - τ) ) and neural networks x(t) - Ax(t) + Bσ(x(t) ) are included as special cases.
领 域: [自动化与计算机技术] [自动化与计算机技术]