Finite horizon state estimation for time-varying neural networks with sensor failure and energy constraint

Abstract

This work studies state estimation for time-varying neural networks (NNs) with sensor failure and energy constraint over finite horizon. The diagonal matrices are introduced to describe the phenomenon of sensor failure, it is then handled by transformed into norm bounded uncertainty. In order to reduce the energy cost, a new transmission strategy is proposed, that is, the measurements firstly transmit using low energy with high packet dropout rate, if the measurements are dropped then the high energy is employed to reduce the packet dropout rate. Sufficient conditions are obtained to ensure that the augmented system satisfies the l2−l∞ performance over finite-horizon, and the estimator gain design algorithm is proposed. At last, a numerical example is shown to verify the derived results.