An efficient automatic modal identification method based on free vibration response and enhanced Empirical Fourier Decomposition technique

This paper presents an efficient yet practical approach for the automatic modal identification of structures based on their free vibration response. The proposed approach relies on the Empirical Fourier Decomposition (EFD) technique. It implements a new procedure to recognize automatically the number of modal components to be extracted from noisy data in such a way to prevent both mode-mixing and mode-splitting effects. A suitable strategy is adopted to improve the segmentation of the frequency spectrum of the free vibration response, so as to identify accurately the bounds of the frequency spectrum partition corresponding to each modal component. The related modal damping ratios are estimated by means of a robust area-based approach in order to mitigate the noise-induced disturbances whereas a time-domain method based on the phase shift of the free vibration response peaks is employed to identify the mode shapes. The proposed approach is first validated through the analysis of synthetic signals that embed closely spaced components and a lowly excited vibration mode. Finally, the proposed approach is applied to two real bridges. The first case-study deals with the identification of modal frequencies and damping ratios of the cables of a stay-cabled bridge. The second case-study involves the modal identification of a steel railway bridge deck that exhibits two closely spaced vibration modes. The outcomes obtained using the proposed approach based on EFD technique are compared with the results obtained by means of the Variational Mode Decomposition (VMD) technique as well as with those computed through classical operational modal analysis techniques. The consistent estimates produced by means of the proposed approach demonstrate its accuracy and robustness.