Nonparametric identification of a particular nonlinear time series system

A nonlinear time series system of cascade structure is identified. The system consists of a nonlinear memoryless element followed by a dynamic linear subsystem. Given a color Gaussian input, the Hermite polynomials based method for recovering the system nonlinearity is presented. The algorithm is carried out by solving an integral equation related to the identification problem. The consistency along with rates of convergence is established. No assumptions concerning continuity of the characteristic or its functional form are required.<
>