Event-triggered state estimation for nonlinear systems aid by machine learning

The event-triggered state estimation problem with the aid of machine learning for nonlinear systems is considered in this paper. First, we develop a recurrent neural network (RNN) model to predict the nonlinear systems. Second, we design a discrete-time dynamic event-triggered mechanism (ETM) and a state observer based on this ETM for the prediction model. This discrete-time dynamic event-triggered state observer significantly reduces the utilization of communication resources. Third, we establish a sufficient condition to ensure that the state observer can robustly estimate the state vector of the RNN model. Finally, we provide an illustrative example to verify the merit of the obtained results.