Analysis of the bias in modal parameters obtained with frequency-domain rational fraction polynomial estimators

Kurzfassung

Modal parameter estimators based on frequency-domain rational fraction polynomials are most commonly used for experimental modal analysis. The Least-Squares Complex Frequency Domain (LSCF) estimator is a well-known member of this group of estimators. However, it has been found out that in particular the damping estimates are biased when the input data is contaminated with noise. Based on closed form analytical equations derived for the Peak-Fit estimator – the simplest single-degree of freedom modal parameter estimator based on frequency-domain rational fraction polynomials – the root cause of this bias is investigated in this paper. The LSCF estimator shows the same kind of bias as the simple Peak-Fit estimator. It is demonstrated for single- and multi-degree of freedom examples that the bias of the LSCF estimator is consistent with the closed form analytical equations of the Peak-Fit estimator. This information can be used to reduce bias of experimental modal analysis results or to improve estimators.