Updating Structural Parameters with Spatially Incomplete Measurements Using Subspace System Identification

Civil infrastructures are subjected to various loads over their lifetime, which leads to structural degradation. To understand and predict these dynamic behaviors of physical infrastructures, many mathematical structural models have been developed, such as subspace system identification. These models often require structural responses from all degrees of freedom (DOFs) to estimate structural parameters. However, in practice, it is often difficult, if not impossible, to make such measurements owing to sensing constraints, a lack of data, or an excessive number of DOFs as in large-scale civil structures. This lack of measurements in space results in ill-posed problems with nonunique solutions. To address this challenge, this paper presents a structural parameter estimation algorithm that incorporates spatially incomplete measurements and inaccurate prior information on structural parameters within a subspace system identification framework. Additional constraints are imposed using prior information, and the prior information is updated with a new estimation. To sequentially update the parameters, the process is repeated as more measurements are collected. The proposed method is evaluated using a numerical model of a 5-story shear building for two damage scenarios with measurement noise. The structural parameters are estimated with 85-99% accuracy with spatially incomplete measurements (40-80%), and the iterative updating further improves these accuracies. (C) 2017 American Society of Civil Engineers.