Identification of linear discrete time systems using linear recurrent neural networks

This paper considers the development of a neural time identification scheme for unknown linear dynamical systems using linear artificial neural networks. The neural networks model used in this paper is a linear recurrent model. The proposed identification scheme is based on minimization of the least squares errors between the actual and the estimated parameters. The analysis and design of this system are discussed. The operating characteristics of the proposed recurrent neural networks for system identification are demonstrated via an example.