8 April 2010 Time-series models for identifying damage location in structural members subjected to ambient vibrations
Author Affiliations +
Abstract
This paper compares two different approaches to identify damage locations in structural members subjected to ambient vibrations. The concept is demonstrated using a simply supported two span steel beam. An electro-hydraulic actuator was used to simulate ambient loading by applying random loads. The vibration time histories were collected for the undamaged and damaged conditions. The structural damages were introduced by cutting notches of different sizes in the flange at different locations. The two different approaches used time-series models in the context of statistical pattern recognition to extract sensitive damage features. In the first method, the damage features were extracted using the errors from fitting autoregressive models with exogenous inputs (ARX) to the collected time histories. The fitted ARX models had been developed based on the undamaged beam. The calculated damage probability from this method could not clearly discriminate the physical damage locations although the change in the condition of the beam was identified. In the second method, variations in the coefficients of multivariate autoregressive models which had been fitted to the acceleration time histories were investigated, and the damage features were extracted by measuring the magnitude of these variations. The findings showed the sensors close to the physical damage locations are related to the larger damage features.
© (2010) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Amir A. Mosavi, Amir A. Mosavi, David Dickey, David Dickey, Rudolf Seracino, Rudolf Seracino, Sami H. Rizkalla, Sami H. Rizkalla, } "Time-series models for identifying damage location in structural members subjected to ambient vibrations", Proc. SPIE 7650, Health Monitoring of Structural and Biological Systems 2010, 76502N (8 April 2010); doi: 10.1117/12.849048; https://doi.org/10.1117/12.849048
PROCEEDINGS
12 PAGES


SHARE
Back to Top