Operational Vision-Based Modal Identification of Structures: A Novel Framework

Abstract

Vision-based modal identification of structures is a rapidly-growing field of study in system identification (SID) and structural health monitoring. However, previously proposed methods typically require many complicated computational steps which limit their application for a fast modal analysis. For instance, the majority of available vision-based methods first convert visual data into vibration time signals and then apply common SID techniques. In the present study, a simple and novel approach is introduced to facilitate a complete modal identification of a structure using widely available tools with no need for data conversion from image to signal. The method is based on Proper Orthogonal Decomposition (POD) which originates from the Karhunen-Loeve expansion and decomposes a set of spatial data measured at various instants in time into a minimum number of orthogonal basis functions using the singular value decomposition. To validate the proposed framework, two experimental case studies are investigated: i) a two-story lab-scale shear frame and (ii) a cantilever with continuous mass. We demonstrate that a set of steps proposed here provide an estimate for modal properties with a high accuracy and resolution.