A multi‐reference‐based mode selection approach for the implementation of NExT–ERA in modal‐based damage detection

SUMMARY A modal‐based global method for identification, localization and quantification of damage of a shear frame structure is presented in this paper. The method involves identification of modal parameters from the vibration responses of the structure only (output only) using a combination of natural excitation technique (NExT) and eigensystem realization algorithm (ERA), and determination of stiffness parameters using very few identified modes by employing a least squares solution procedure. Finally, damage is indicated by comparing the stiffnesses of the (possibly) damaged structure with that of the undamaged structure. The implementation of NExT–ERA requires a reference channel for modal identification, but several limitations of using a single reference channel have been identified from the investigations of this study. To overcome the limitations, a new approach called multi‐reference‐based mode selection is proposed in this paper. The approach makes use of all the available DOFs, but one at a time, as reference channels and identifies the best representative modes of the structure by a mode screening process and a mode separation process. The approach is also automated for expediency and possible real‐time application. The presented damage detection methodology is implemented on the analytical phase I and experimental phase II of the IASC‐ASCE SHM benchmark problem and shows an accuracy of around 90% in damage detection. Copyright © 2014 John Wiley & Sons, Ltd.