Nonlinear System Modeling Based on Non-Parametric Identification and Linear Wavelet stimation of SDP Models

This paper describes a data-based approach to the identification and estimation of nonlinear dynamic systems which exploits the concept of the state dependent parameter (SDP) model structure. The major attractive features of the proposed approach are the initial identification of the nonlinear system's structure that analyzes the nature of the associated nonlinearities by the non-parametric estimation of the SDP model using recursive fixed interval smoothing; and a compact parameterization of this initially identified parsimonious model structure via a linear wavelet functional approximation, prior to final parametric optimization. A simulation example is used to demonstrate the proposed approach