Mixed H[formula omitted] and passivity finite-time state estimation for neural networks under hybrid cyber-attacks

Abstract

This paper focuses on finite-time state estimation for neural networks under hybrid cyber-attacks based on adaptive event-triggered scheme (AETS). Firstly, Markov jump process is used to describe different types of cyber-attacks. In order to save communication resources, the system output state model under mixed cyber-attacks is established based on AETS. At the same time, a state estimator based on the output signal is designed to avoid the state unavailability of the system. By constructing a Lyapunov-Krasovskii functional and affine Bessel-Legendre inequality method, sufficient conditions are obtained to guarantee the estimation error system is finite-time boundedness and mixed passive and H∞ performance level. The expected gain of the state estimator is obtained by solving a set of feasible linear matrix inequalities. Finally, the robustness and effectiveness of the proposed method are verified by three numerical examples such as Chua’s circuit.