一类具有混合时滞的离散型递归神经网络稳定性分析

Stability analysis on discrete-time recurrent neural networks with mixed delays

为了进一步降低具有混合时滞离散型递归神经网络系统稳定性判据的保守性,建立了1个改进的基于自由权矩阵的双重求和不等式,通过构造1个新的含更多激励函数信息的Lyapunov-Krasovskii泛函,结合Jesen不等式和2个改进的求和不等式估计所构造Lyapunov-Krasovskii泛函前向差分的上界,获得了保守性更低的一类具有混合时滞的离散型递归神经网络系统全局渐近稳定性判定条件,并结合数值实例验证了所得结论的有效性和优越性。

In order to further reduce the conservatism of stability criteria for discrete-time recurrent neural networks with mixed delays,an improved double summation inequality based on free-weighting matrices is proposed.,and a new Lyapunov-Krasovskii functional with more information on activation function is defined.When estimating the difference of the considered Lyapunov-Krasovskii functional,the Jensen inequality and two improved summation inequalities are introduced.Then a new asymptotic stability criterion with less conservative is obtained.Finally,one numerical example is presented to demonstrate the effectiveness and superiority of the result.