Output-only modal parameter identification of structures by vision modal analysis

Abstract

Testing of structures by non-contact and possibly-remote video camera measurement has emerged as a promising topic in engineering fields. This paper develops a vision modal analysis framework for output-only structural modal parameter identification using vision measurement. The establishment of this framework is mainly twofold. On the one hand, the vision data as the intensities of some selected pixels on a vibrating structure is elaborately derived upon the optical flow theory to possess the vision modal decomposition. Under small motion and proper smoothness assumptions, such a decomposition is similar to the conventional structural modal decomposition, indicating that the structural modal parameters are directly implicated in the vision data. While in the case of finite motion or a non-smoothing video, additional higher-order harmonics as combinations of basic frequencies may arise. On the other hand, with the vision modal decomposition, the frequency domain decomposition is utilized to identify the vision modal parameters from the vision data and then, the structural mode shapes are recovered from the vision modal shapes by some motion extraction processing. Numerical examples and an experimental case are studied to see the feasibility and accuracy of the proposed vision modal analysis.