Further improved results on non-fragile H∞ performance state estimation for delayed static neural networks

Abstract

This work investigates the non-fragile H∞ state estimation issue for static neural networks (SNNs) with mixed time-varing delays and randomly occurring uncertainties (ROUs). ROUs with Bernoulli distributed white noise sequences are firstly considered for tackling state estimation of SNNs, which are mutually uncorrelated stochastic variables. In order to take full advantage of the slope information about activation function (SIAAF), the estimation error of activation function is separated into two parts. Based on the more SIAAF, a modified Lyapunov-Krasovskii functional (LKF) is constructed. In addition, by combining integral inequality and zero equality with several parameters, further improved results are emerged to ensure the error system is globally asymptotically stable with a prescribed level γ and reduce conservativeness to some extent. Furthermore, the more practical non-fragile estimator gain matrix can be obtained via the above designed optimization algorithm. Finally, two numerical examples are furnished to verify the effectiveness and performance of the developed method.