Abstract

The primary objective of Structural Health Monitoring (SHM) is to diagnose structures for damage, take necessary measures if any damage occurs, and estimate their degradation rate. Conventional non-destructive evaluation methods are not always practical for implementation of a continuous health monitoring system. Vibration Based Damage Identification (VBDI) methods applied to SHM can be useful in interpreting the global vibration response of a structure to identify local changes. Due to complicated features of real life structures there are some uncertainties related to input parameters such as measured frequencies and mode shape data, where output is sensitive to errors in modal parameters. As all VBDI processes rely on experimental data with their inherent uncertainties, statistical procedures are helpful if one is to interpret the vibration response mixed with other ambient affects. The objective of this study is the detection of damage by VBDI methods and statistical pattern recognition techniques. Here, two practical structures, the Crowchild Bridge in Calgary, and a 3D-Space Frame have been tested with two VBDI algorithms. The Damage Index and Matrix Update methods have been selected to study simulated damage cases on the numerical models of the selected structures. For the application of statistical pattern recognition techniques to damage identification, another in-service structure, the Portage Creek Bridge in Victoria, Canada has been tested. The classification of the patterns has been performed using outlier analysis. Alternatively, damage detection by pattern comparison using residual errors has been applied.