Nonlinear system identification in a noisy environment using wavelet based SDP models

This paper presents an approach to the identification of nonlinear system in noisy environment using a wavelet based State Dependent Parameter (SDP) model to chacterize the system's nonlinear dynamics. The obtained model is in the form of a set of linear regressive output/input terms (state) multiplied by the respective SDPs, which are compactly parameterized by wavelet basis functions. In this approach,a modified Instrumental Variable (IV) algorithm is used to solve to the inconsistency problem of linear least squares parameter's estimation in the presence of noise. A simulation example is provided to illustrate the proposed approach.