Output-Only Modal Identification Based on Auto-regressive Spectrum-Guided Symplectic Geometry Mode Decomposition

Purpose The purpose of this paper is to propose a novel output-only structural modal identification approach based on an improved symplectic geometry mode decomposition (SGMD) method, i.e., auto-regressive (AR) spectrum-guided symplectic geometry mode decomposition (ARSGMD). Therefore, output-only modal identification for structures based on the signal decomposition technique is studied. Methods First, we improved the SGMD by introducing the AR spectrum to determine the number of iterations and the frequency bound for each iteration of SGMD, and the dynamic response is decomposed into several symplectic geometry components (SGCs) containing one major frequency by the ARSGMD. Then, the free decay response (FDR) can be obtained by the random decrement technique (RDT) and the mode shape can be extracted by the modal responses extracted from the FDRs at all available sensors. Finally, the natural frequencies and damping ratio can be estimated by the Hilbert transformation (HT). Results The performance and superiority of the proposed ARSGMD method is validated by numerical example and compare with others methods. The result shows that the ARSGMD has better ability in signal decomposition for structural modal identification. The feasibility of the proposed ARSGMD-based output-only modal identification approach is demonstrated by the numerical example and experimental investigation. The results show that the proposed method can identify the modal parameters accurately. Conclusions Associate with the AR spectrum, the proposed ARSGMD approach can decompose the complex signal into several single frequency components adaptively without mode mixing and over-decomposition, even for noise-contaminated signal. The proposed ARSGMD-based modal identification method can effectively extract the structural modal parameters under impulse or ambient excitation. Besides, the closely spaced modes can be accurately decomposed and extracted through the proposed ARSGMD-based modal identification approach.