AMD-Based Random Decrement Technique for Modal Identification of Structures with Close Modes

Abstract

Random decrement technique (RDT) is a popular time-domain approach to extract modal properties of structures from ambient vibration data; however, it may result in poor estimation results when structural modes are closely spaced. In this study, a method of combining analytical mode decomposition (AMD) and RDT is presented to determine the modal properties of structures with closely spaced modes from ambient vibration data. The measurement acceleration data are first decomposed into a series of subsignals by way of the AMD. Then, the RDT is applied to each subsignal to extract the random decrement signature from which the modal properties of the structure are identified. The proposed AMD-based RDT method is compared with the multimode random decrement technique (MRDT) and stochastic subspace identification (SSI) through numerical simulation data from a four-degrees-of-freedom system with close modes. It is shown that the present method performs better than the MRDT and SSI. When significant modal interaction occurs, decomposing into the multimode subsignals successfully separates responses of close modes from those of other modes, which permits accurate identification of the modal properties for the relevant modes. The modal parameters of a curved cable-stayed footbridge are estimated by the proposed method, demonstrating that the method is viable in practical applications.