System Identification of Dynamic Characteristics with Ambient Vibration Measurements

摘要

A novel approach for system identification of dynamic characteristics based on ambient vibration measurements is proposed in the paper for structures with closely spaced modes. The time series analysis is often adopted for identifying structural dynamic characteristics based on ambient vibration. However, the problems caused by closely spaced modes and spurious modes contradict its application to civil infrastructures. In the case of free vibration, which is equivalent to impulse function, the magnitudes of all eigen-values in the conventional Auto-Regressive (AR) model are less than one. On the contrary, for the Backward Auto-Regressive (BAR) model, the magnitudes of all eigen-values of system modes are less than one. Based on the principal, the system modes are determined by the magnitudes of the eigen-values of the characteristic polynomial. By using of the equivalent relationship between correlation function and impulse response function, a Correlation Function Backward Auto-Regressive model (CFBAR) was formulated. The theoretical derivation demonstrates the proposed method has the noise-free characteristic. The feasibility and effectiveness of the proposed method for modal identification based on ambient vibration measurements was validated by numerical simulation.