Bayesian operational modal analysis with buried modes

Abstract

In full-scale ambient vibration tests, challenging situations exist where in the frequency domain the measured data is dominated by other modes that ‘bury’ the subject mode of interest. In this case, conventional modal identification methods are either not applicable or inefficient to apply. This paper proposes a Bayesian frequency domain method for identifying the modal properties of such buried modes. The buried-mode situation is modelled and computation difficulties are addressed, leading to an efficient algorithm for modal identification in such challenging situation. The proposed method is validated by synthetic data examples. The associated uncertainty of the identified modal parameters are investigated. The method is also applied to identifying the buried modes of a long-span suspension bridge, demonstrating its utility with challenging modes encountered in field test data.