Application of image - based sensing methods in shake table experiments

Abstract

Earthquake Engineering-related experimental dynamic testing of Single Degree of Freedom (SDOF) or Multi Degree of Freedom (MDOF) structural systems involve high precision measurement devices such as strain gauges, Linear Variable Displace ment Transducers (LVDTs), Linear Potentiometer (LPT), lasermeters and accelerom eters. Also, high frequency data acquisition systems need to be used to collect data within a narrow time frame. Higher sensitivity and accuracy mean higher measure ment costs. Thus, they are not cost effective, and this becomes a major disadvantage in dynamic shake table experiments. However, image-based sensing methods could present a feasible alternative way to perform such measurements. This study inves tigates the applicability of image-based methods in various forms of dynamic exper iments of structural specimens. In the past, similar studies have been completed on this topic. Displacements of different structures under ambient vibrations or earth quake ground motions were investigated with different tracking methods. This study mainly focuses on two different vision-based methods to extract displacement data within short-period and high frequency dynamic motion. One method is a simple LED light source tracking, while the other one uses an existing object tracking algorithm (Kanade-Lukas-Tomasi). Quality of camera in terms of resolution and image capturing capability (frames per second) coupled with the image algorithms were compared. A computer program with an appropriate user interface was developed to analyze the video images. Displacement, velocity, acceleration and strain data provided by image based measurements were compared with the results from conventional devices such as strain gauges, accelerometers and LVDT’s.