Finite-Horizon $H_{\infty }$ State Estimation for Discrete Time-Varying Artificial Neural Networks: An Accumulation-Based Event-Triggered Mechanism

This paper investigates the finite-horizon $H_{\infty }$ state estimation problem for a class of discrete time-varying artificial neural networks. In order to reduce energy consumption, a novel accumulation-based event-triggered mechanism (ABETM) is proposed to regulate the signal transmissions between the network outputs and the estimator. Compared with the traditional event-triggered schemes, the ABETM is perceived to be more robust against unexpectedly abrupt changes of the signals that could occur due to measurement noises, sampling problems and unusual conditions. The objective of this paper is to design an $H_{\infty }$ state estimator such that the error dynamics of the state estimation satisfies the prescribed $H_{\infty }$ performance requirement on a finite horizon. A sufficient condition is first provided to guarantee the finite-horizon $H_{\infty }$ performance of the state estimator by resorting to the Lyapunov stability theory. Then, the desired estimator gain is calculated by solving some matrix inequalities. Finally, a simulation example is given to verify the validity of the proposed estimator under the ABETM, where the advantages of the proposed ABETM are demonstrated as compared to the traditional event-triggered schemes.