Study On Closely Spaced Modes Decomposition And Modal Parameter Identification

Modal parameter identification of vibration is required by structural vibration characteristics analysis, damage diagnosis and forecast, and optimization design of structural dynamics characteristics. The time-frequency joint analysis method, which developed from information engineering, can analyze the time-frequency characteristics of signals in both time and frequency domains. It's a powerful tool for non-stationary signals, and made a great contribution to the development of experimental modal analysis. Based on the summarization of research and application of wavelet transform and Hilbert-Huang transform in modal parameter identification, a closely spaced modes decomposition method is proposed to improve the precision of modal parameter identification of structures with closely spaced modes.The reasons for low precision in modal parameter identification of closely spaced modes are analyzed. Considering the two main factors, natural frequency and damping ratio, which affect the modes overlap seriously, a factor so-called modes closely degree is defined to evaluate modes overlap. A method based on positive exponential window and band-pass filter is proposed to decompose closely spaced modes. Furthermore, noise reduction methods are researched. It's proved that the parameters won't be changed after correlation processing if the damping ratios are small enough.The modal parameters of a 4-DOF(degree-of-freedom) model and a 3-DOF model with closely spaced modes are identified respectively using wavelet transform. The precision of identification results of closely spaced modes are poorer than others obviously. Then, the closely spaced modes decomposition method is applied to the 3-DOF model to improve it.The modal parameters of a 4-DOF model and a 3-DOF model with closely spaced modes are identified respectively using Hilbert-Huang transform. The precision of identification results of closely spaced modes are poorer than others obviously. Then, the closely spaced modes decomposition method is applied to the 3-DOF model to improve it.Analyzing the experimental modal data of the bridge over Xining Beichuan River based on band-pass filtering, the modal parameters are identified by wavelet transform and Hilbert-Huang transform respectively. Compared to the results in the reference, the band-pass filtering is proved to be effective. Moreover, a 2-DOF experimental model with closely spaced modes is designed, and the testing data of a building is considered too. The closely spaced modes decomposition method is testified to be effective on promoting identification Accuracy.