Frequency Response Function Measurements via Local Rational Modeling, Revisited

A finite measurement time is at the origin of transient (sometimes called leakage) errors in nonparametric frequency response function (FRF) estimation. If FRF varies significantly over resolution experiment (= reciprocal time), then these transients (leakage) cause important bias and variance estimate. To decrease errors, several local modeling techniques have been proposed literature. This article presents an overview existing methods gives in-depth analysis disturbing noise estimates. In addition, a new approach described, which combines small error rational approximation with low sensitivity polynomial approximation. It based on automatic model-order selection procedure applied to specific subclass functions.