H∞ State Estimation for Neural Networks Subject to Missing Measurements with Uncertain Missing Probabilities*

This paper addresses the H∞ state estimation problem for a type of neural networks with missing measurements. A series of stochastic variables obeying Bernoulli distributions is adopted to characterize the randomly occurring missing measurements, where the missing probability is considered to be changeable over time and satisfies a norm-bounded condition. This paper aims to construct an effective state estimator such that the mean-square asymptotical stability and the specified Н∞ performance of the augmented estimation error system can be achieved. A criterion is given for the existence of the admissible estimator according to a combination of Lyapunov theory as well as stochastic analysis technique, and the estimator parameters are obtained by dealing with a convex optimization problem. Finally, a simulation example is carried out to illustrate the usefulness of the provided estimation scheme.