数学物理学报(英文版) ›› 2019, Vol. 39 ›› Issue (1): 195-213.doi: 10.1007/s10473-019-0116-7

• 论文 • 上一篇    下一篇

SAMPLED-DATA STATE ESTIMATION FOR NEURAL NETWORKS WITH ADDITIVE TIME–VARYING DELAYS

M. SYED ALI1, N. GUNASEKARAN1,2, 曹进德3   

  1. 1. Department of Mathematics, Thiruvalluvar University, Vellore, Tamilnadu, India, 632 115, India;
    2. Research Center for Wind Energy Systems, Kunsan National University, Kunsan, Chonbuk, 573-701, Korea;
    3. School of Mathematics, Southeast University, Nanjing 210096, China
  • 收稿日期:2017-09-07 修回日期:2017-12-14 出版日期:2019-02-25 发布日期:2019-11-14
  • 通讯作者: M. SYED ALI E-mail:syedgru@gmail.com
  • 作者简介:N. GUNASEKARAN,E-mail:gunasmaths@gmail.com;Jinde CAO,E-mail:jdcao@seu.edu.cn

SAMPLED-DATA STATE ESTIMATION FOR NEURAL NETWORKS WITH ADDITIVE TIME–VARYING DELAYS

M. SYED ALI1, N. GUNASEKARAN1,2, Jinde CAO3   

  1. 1. Department of Mathematics, Thiruvalluvar University, Vellore, Tamilnadu, India, 632 115, India;
    2. Research Center for Wind Energy Systems, Kunsan National University, Kunsan, Chonbuk, 573-701, Korea;
    3. School of Mathematics, Southeast University, Nanjing 210096, China
  • Received:2017-09-07 Revised:2017-12-14 Online:2019-02-25 Published:2019-11-14
  • Contact: M. SYED ALI E-mail:syedgru@gmail.com

摘要: In this paper, we consider the problem of delay-dependent stability for state estimation of neural networks with two additive time-varying delay components via sampleddata control. By constructing a suitable Lyapunov-Krasovskii functional with triple and four integral terms and by using Jensen's inequality, a new delay-dependent stability criterion is derived in terms of linear matrix inequalities (LMIs) to ensure the asymptotic stability of the equilibrium point of the considered neural networks. Instead of the continuous measurement, the sampled measurement is used to estimate the neuron states, and a sampled-data estimator is constructed. Due to the delay-dependent method, a significant source of conservativeness that could be further reduced lies in the calculation of the time-derivative of the Lyapunov functional. The relationship between the time-varying delay and its upper bound is taken into account when estimating the upper bound of the derivative of Lyapunov functional. As a result, some less conservative stability criteria are established for systems with two successive delay components. Finally, numerical example is given to show the superiority of proposed method.

关键词: Lyapunov method, linear matrix inequality, state estimation, sample-data control, time-varying delays

Abstract: In this paper, we consider the problem of delay-dependent stability for state estimation of neural networks with two additive time-varying delay components via sampleddata control. By constructing a suitable Lyapunov-Krasovskii functional with triple and four integral terms and by using Jensen's inequality, a new delay-dependent stability criterion is derived in terms of linear matrix inequalities (LMIs) to ensure the asymptotic stability of the equilibrium point of the considered neural networks. Instead of the continuous measurement, the sampled measurement is used to estimate the neuron states, and a sampled-data estimator is constructed. Due to the delay-dependent method, a significant source of conservativeness that could be further reduced lies in the calculation of the time-derivative of the Lyapunov functional. The relationship between the time-varying delay and its upper bound is taken into account when estimating the upper bound of the derivative of Lyapunov functional. As a result, some less conservative stability criteria are established for systems with two successive delay components. Finally, numerical example is given to show the superiority of proposed method.

Key words: Lyapunov method, linear matrix inequality, state estimation, sample-data control, time-varying delays