H∞ State Estimation for Time-Delayed Static Neural Networks Using an Improved Delay-Product-Type LKF

This paper is concerned with
$H_{\infty}$
performance state estimation of static neural networks with a time-varying delay. First, a PI estimator with exponential term is used to estimate neuron states based on output measurement. Second, an augmented Lyapunov-Krasovskii functional (LKF) containing delay-product-type non-integral terms and single integral terms is constructed by introducing negative definite terms. After that, a criterion with less conservatism is derived based on extended reciprocally convex matrix inequality. Finally, a numerical example is provided to reveal the effectiveness of the proposed approach.