Enhanced EMD-RDT Method for Output-Only Ambient Modal Identification of Structures

Abstract

An enhanced approach of empirical mode decomposition based random decrement technique (EMD-RDT) is proposed for output-only modal parameter identification of structures using ambient vibration measurements. In the conventional EMD-RDT method, modal parameters are identified for each mode through Hilbert transform or least-square fitting from the free-decay modal responses, which are produced by EMD and RDT sequentially. The identification process is time consuming, and many uncertainties are involved. The novel enhancements of the proposed method lie in two aspects: computation efficiency and uncertainty treatment. On one hand, a novel decomposition method is proposed to separate the mode shape and the modal coordinates from the free-decay modal responses, thus making the identification process more efficient. On the other hand, a bootstrap approach is employed to quantify the uncertainties of modal parameters by providing surrogate estimates to generate useful statistics. Examples using both simulated data from a six degrees-of-freedom (six-DOF) system and experimental data from a three-story shear building structure are presented to demonstrate the proposed method, of which the effectiveness and the efficiency are confirmed by the identified results.