Assessing uncertainty bounds in frequency domain decomposition

Frequency Domain Decomposition (FDD) appears as the most preferred one among the Operational modal analysis (OMA) techniques. Although the underlying theory of FDD and its modified versions are well stated in the literature, the uncertainty bounds on the modal parameter estimates have not been investigated in detail. To address this problem, an analytical approach is presented in this study. In this context, first the mode shape covariance matrix is estimated via an augmented eigenvalue problem through the perturbation approach. Then, the covariance matrix for modal frequencies and damping ratios are derived based on a nonlinear regression problem using Fisher information. To assess the effects of the presented approach, one numerical and one experimental study are conducted. The achieved results reveal that the proposed approach can make successful estimates for the uncertainty propagation in FDD.