State estimation for recurrent neural networks with unknown delays: A robust analysis approach

In this paper, the state estimation problem is investigated for recurrent neural networks with unknown delays both in state equation and output equation. By constructing the Taylor series and linear matrix inequality (LMI) technique, the sufficient conditions of state estimation for a kind of recurrent neural networks with unknown delays are presented, therefore, the error system is globally asymptotically stable with L∞ performance. The design observer can attenuate the effect of the unknown delays on a pre-defined performance output, and the observer gain can be obtained by solving a set of linear matrix inequalities. Some remarks and examples are given to show the effectiveness of the proposed method in comparison with some existing results.