不确定时滞动态反馈神经网络的鲁棒分析与设计

Robust Designing for Dynamic Feed-back Neural Networks With Structural Uncertainties and Time-Delay

研究了一类具有时变时滞和结构不确定性的Hopfield神经网络鲁棒稳定性问题 ,应用线性矩阵不等式 (LMI)方法得到了不确定时变时滞神经网络全局鲁棒稳定的充分条件 ,特别是在给定网络平衡点允许偏移的情况下 ,通过验证定理中给出的LMI,可以非常方便地进行神经网络的鲁棒设计。数值仿真表明了方法的有效性。

Robustness of a kind of dynamic feed-back neural networks with structural uncertainties and time-variable delay is investigated and an useful theorem on the global robust asymptotic stability is deduced by applying LMI method. Under the permitted deviation of the equilibrium it can be designed. The result will be very convenient for us to design a NN with excellent robustness.