| Resumo: | The development of efficient vibration based structural health monitoring systems requires distinguishing between abnormal changes in modal parameters caused by structural damage and normal changes due to varying environmental conditions. In this context, this paper is focused on the analysis and extraction of effects of temperature oscillations on natural frequencies of a footbridge, where a long-term dynamic monitoring system was installed by the Laboratory of Vibrations and Monitoring of the Faculty of Engineering of University of Porto. With that purpose, firstly, the temperature influence on the natural frequencies is reported, and correlations between measured temperatures and estimated natural frequencies are analyzed. Then, the Principal Component Analysis (PCA) and the Novelty Detection method are applied to identified natural frequencies: PCA effectively eliminates environmental influence; Novelty analysis on the residual error of PCA predicted model is used as a statistical indication of damage. The proposed procedure is illustrated using continuous dynamic data collected from the footbridge during more than one year.
|
|---|