Structural health monitoring (SHM) is becoming a more widely accepted tool to improve bridge management. This paper describes a fiber optic strain-based SHM system which was developed to continuously monitor bridge performance under ambient traffic loads to detect damage. Strain time history data collected by this system are utilized to construct a baseline model that is based upon the linear prediction models obtained from extreme event strains of sensor pairs. Residuals are then calculated as the difference between the measured strains and the predicted strains calculated by assuming the structure is undamaged. Residuals deviating from their baseline distributions, therefore, are indicators of damage or degradation. Control chart analysis is utilized to identify/define the statistically significant deviation. In order to separate the operational and environmental factors that are known to induce strain variance from those resulting from structural condition changes, various data pre-processing algorithms are utilized.Data zeroing and filtering approaches eliminate temperature induced strain and truck/bridge vibration effect, respectively. A specially designed strain based truck detection sub-system was developed to support a live loadcondition selection function, so that live load variances (e.g., vehicle type, vehicle geometry, vehicle positioning, etc.) can be minimized. The effectiveness of the system was demonstrated with field collected pre-damage data and synthetic post-damage data.
Samenvatting