Ambient Vibration Analysis with Subspace Methods and Automated Mode Selection: Case Studies

Abstract

This study presents an investigation of the performance of subspace techniques for modal identification using ambient vibration measurements. Several models and structures characterized by increasing degrees of complexity are investigated to assess the potential benefits of stochastic subspace identification algorithms, and the difficulties that might be experienced during a modal identification analysis. The case studies include a simple three-degree-of-freedom 3 DOF mass-spring-dashpot model, the 120 DOF finite element model, as well as the physical laboratory model of a small scale steel frame, and a long span suspension bridge. Stabilization diagram and clustering analysis approaches are adopted for spurious mode rejection, and the latter is found to be promising for automating the mode selection process. The experiences with experimental data reveal that some preconditioning tools are quite helpful in order to properly focus on the structural modes of interest, and that preconditioning improves the performance of the subspace methods. On the whole, the approaches investigated in this study are found to perform quite satisfactorily for operational modal analysis of engineering structures.