本文研究一类具有泄漏时滞的双向联想记忆（BAM）神经网络的渐近接近问题。首先，通过适当的等价转换，将神经网络系统转换为中立型时滞系统。然后，利用合适的Lyapunov泛函和不等式方法，得到了时变输入系统的任意解之间渐近接近、以及此系统的解渐近接近其相应的常输入系统平衡解的判别准则。本文所得结论在一定程度上推广了Gopalsamy在Journal of Mathematical analysis and Applications [325(2007):1117-1132]上的研究工作。
This paper is concerned with the asymptotic approximation of a class of bidirectional associative memory (BAM) neural networks with leakage delays. First, the systems are written as neutral delayed systems by a proper transformation. Then, by using appropriate Lyapunov functionals and an inequality method, a couple of criteria are obtained, which can guarantee the interior asymptotic behaviors of our BAM neural network with time-varying inputs, and the asymptotic behaviors between the BAM neural network and its corresponding one with constant inputs. The present results we obtained generalize to some extent the works published in the Journal of Mathematical analysis and Applications [(325) 2007: 1117--1132].
Key words： BAM Neural Network; Leakage Delay; Lyapunov Function; Asymptotic Approximation