New Stability Criteria for Recurrent Neural Networks with Leakage Delay and Time-varying Interval Delay
摘要:
研究了具有漏泄时滞和时变区间传输时滞的递归神经网络的渐近稳定性问题.基于Lyapunov-Krasovskii(L-K)稳定性理论.Jensen不等式和互惠凸方法,得到了线性矩阵不等式(LMIs)表示的新的稳定性准则.相对于现存的方法,避免利用保守性较大的中立型变换,且在构造L-K泛函时充分利用了漏泄时滞和传输时滞的关联信息,因此所得准则具有较小的保守性.数值例子验证了所得结果的有效性和较小保守性.
This paper considers the asymptotic stability problem for recurrent neural networks with leakage delay and time-varying interval transmission delay. Based on Lyapunov-Krasovskii (L-K) stability theory, Jensen inequality, and recipro- cally convex approach approach, new asymptotic stability criteria are obtained in terms of linear matrix inequalities (LMIs). Compared with the existing approaches, this paper avoids using the neutral transformation, and utilizes the interconnected in- formation between leakage delay and transmission delay sufficiently when constructing L-K functionals, thus the obtained crite- ria are less conservative. Numerical example verifies the effectiveness and less conservatism of the obtained results.
作者:
王军涛 郑群珍 苏展 陈永刚
机构地区:
河南科技学院数学科学学院 河南教育学院数学与统计学院 上海辅仁医药研发有限公司
出处:
《betway官方app 学报:自然科学版》 CAS 北大核心 2016年第3期166-171,177,共7页
基金:
国家自然科学基金(61304061)
关键词:
递归神经网络 漏泄时滞 时变时滞 渐近稳定性
recurrent neural network leakage delay time-varying delay asymptotic stability
分类号:
TP183 [自动化与计算机技术—控制理论与控制工程]