A moving‐window least squares fitting method for crack detection and rigidity identification of multispan bridges

SUMMARY In this study, a moving‐window least squares fitting method is proposed for rapid identification of cracks and flexural rigidities in multispan bridges. First, the dynamic deflections of a continuous bridge were locally measured under a dynamic point load. Their integrations over time, referred to as ‘integration‐over‐time deflections’, were used to derive ‘integration‐over‐time slopes’. These virtually static measurements over a short segment of the bridge were then fitted into a cubic curve in the least squares sense. Finally, the coefficient of the square term of the fitted curve was used to determine both the magnitude and location of local flexibility because of cracking and/or changing in flexural rigidity of the bridge. For multispan continuous bridges, an iterative procedure was developed to ensure that the end moments of various spans are compatible with the identified cracks and rigidity changes. To illustrate the proposed method, prismatic girder bridges with multiple cracks of various depths or non‐prismatic girder bridges were analyzed. Sensitivity analysis was conducted on the effects of weighting factor, noise level, load type, window length, and bridge discretization. Numerical results demonstrated that the proposed method can accurately detect cracks and identify the change in flexural rigidity. The five‐point equally weighted algorithm is recommended for practical applications. The spacing of two discernible cracks is equal to the window length. The identified results are insensitive to noise because of integration of the dynamic measurements. Copyright © 2011 John Wiley & Sons, Ltd.