一类具有混合时滞神经网络的渐近稳定性判据

A Criterion on the Global Asymptotical Stability of the Neural Network with Continuous Time-Varying and Distributed Time Delays

摘要:

在激励函数仅仅满足扇区条件,时滞函数连续可微且导数小于0的情形下,通过构造Lyapunov- Krasovskii泛函,得到了神经网络全局渐近稳定的新判据。该判据可以通过线性矩阵不等式(LMI)表达,借助解LMI的内点法可以方便求解。

Abstract:

Under the condition that the activation function are only monotonically nondiscreasing and time delay function is continuously differentiable, its derivative is smaller than 0, a new criterion for guarantying the equilibrium point of the neural network be global asymptotically stable is derived in this paper by constructing a appropriate Lyapunov-Krasovskii functional. The resulting criterion is described by the LMI; and the solution can be easily obtained via the existing numerical algorithms such as the interior-point algorithms.