State estimation for delayed neural networks with stochastic communication protocol: The finite-time case

This paper is concerned with the finite-time state estimation problem for a class of delayed artificial neural networks under the stochastic communication protocol. The underlying time delay is time-varying yet bounded. Compared with the common-used sigmoid-type nonlinearity, a more general type of nonlinearity is adopted to describe the neuron activation function and the nonlinearity of the measurement output, respectively. In order to avoid the communication collision, the stochastic communication protocol is introduced between the transmitter and the receiver, and the corresponding scheme is characterised with the help of a Markov chain. By introducing an auxiliary vector, a novel state estimator structure is proposed. The stochastic finite-time stability of the error dynamics is first analyzed via the stochastic analysis techniques and the Lyapunov stability theory, and then the sufficient condition for the existence of the desired state estimator is obtained. Subsequently, the estimator gain is parameterized by using a set of easy-to-check computational condition. Finally, a numerical example is provided to show the effectiveness of the proposed algorithm.