Shift data to distinguish spurious modes in eigensystem realization algorithm

Abstract

Modal parameters, such as frequency, damping ratio and modal shape, play a key role to reflect the structural behaviour. Eigensystem realization algorithm (ERA) has been effectively utilized to many applications to identify the modal parameters. However, the noises in practical environment can influent the effectiveness of ERA, which can introduce spurious modes. This paper proposes a modified ERA to distinguish the spurious modes. It is first proved that any two Hankel matrices with one time step shift used in ERA can obtain the state matrix. The singular values in the singular value matrix obtained by singular value decomposition of Hankel matrix are populated by noises. The truncated order for the singular value matrix cannot be determined. It is proposed that singular value changing can be observed through moving data, which help to identify the spurious mode. The stabilization diagram is employed to obtain the interested modes. Then the proposed moving data diagram can demonstrate the singular value changing which is spurious. Finally, the performance of the proposed method is verified through a numerical example.