System identification of linear structures using Hilbert transform and empirical mode decomposition

In this paper, simple procedures based on the Hilbert transform and Empirical Mode Decomposition are proposed to identify MDOF linear structural systems using the measured impulse response time histories. Measured response data are first decomposed into Intrinsic Mode Functions using the Empirical Mode Decomposition method. These Intrinsic Mode Functions are shown to be the modal responses. Then, the Hilbert transform is applied to each Intrinsic Mode Function to obtain the amplitude and phase angle time histories of each mode, from which natural frequencies, damping ratios and mode shapes as well as the physical mass, stiffness and proportional damping matrices of the structure are identified. The applications of the methodology presented are demonstrated through numerical simulations. Simulation results indicate that the system identification method presented offers a simple and effective tool for parametric identification of linear structures.