FDD based modal identification of structures using least squares approach

Abstract

In the conventional frequency domain modal identification procedures, modal parameters are attempted to be identified using power spectral density matrix decomposition. As it is well stated in the literature, the quality of estimated mode shapes significantly increases when an average PSD matrix over a selected frequency band is utilized in the identification procedure. In this context, this study aims at deriving a weighted average for the PSD matrix to increase the quality of the estimated mode shapes by using an FDD-based modal model. Then, the remaining modal parameters comprising modal frequency, damping ratio and spectral density of modal exciting force are calculated through a least squares equation. Moreover, the presented FDD based approach is also adapted to the solution of multiple setup problem. The proposed method for single and multiple setup cases is validated by making use of one numerical and two field data examples. The obtained results reveal that the presented frequency domain technique can successfully achieve high quality of modal estimates in the presence of measurement noise, and it can be easily implemented to field applications.