Neural Network Based State Estimation of Dynamical Systems

Abstract

A neural network based state estimator for a general class of nonlinear dynamic system is proposed. The proposed state estimator uses cascading of a recurrent neural network structure (RNN) which learns the internal behavior of the dynamical system and a feedforward neural network (RNN) which learns the measuring relations of the system from the input-output data through prediction error minimization. A dynamic learning algorithm for training the recurrent neural network has been developed. The proposed method has been evaluated with different applications.