System identification by using adaptive Fourier decomposition method with polynomials

Abstract

Adaptive Fourier decomposition (AFD) method was used in system identification in recent years. This algorithm is a greedy technique of approximation, it is to find fast approximations by selecting basis functions. In this paper, a continue work is presented to study the two-step algorithm developed for identification of linear time-invariant systems by using AFD algorithm. The first step in the two-step algorithm is modified, where polynomials are used to replace Cauchy formula. Analysis shows the robustness is kept. The effectiveness of new improvement is shown through an example.