Local damage detection of bridges using signals of strain measurements

An energy-based local damage detection technique for long-span railway bridges is presented. The philosophy is that any change in structural continuity, material property and connection between different components would reflect an alteration in effective stiffness and thus lead to the variation of energy distribution in the structure. By investigating the energy behaviour, the health state of structure can be evaluated, especially in the case of external excitations having consistent intensity. Field measurements indicate that railway traffic induced vibration may dominate dynamic response of long-span bridges and provides an excitation with statistical uniformity. Measured strain time-history sequences are used for collecting signal segments containing railway traffic induced vibration for the analysis of strain energy and segmental statistics. The results are then utilized to extract pattern features and construct statistical parameters. The bootstrap method is advocated as a practical tool for estimating confidence intervals of the extracted parameters. It makes the evaluation procedure to be performed quantitatively and easily.