Updating Structural Parameters with Spatially Incomplete Measurements Using Subspace System Identification

Abstract

Civil infrastructures are exposed to various loads over their lifetime, which lead tostructural degradation. To understand and predict these dynamic behaviors of thephysical infrastructure, many mathematical structural models have been developed,such as subspace system identification. These models often require structuralresponses from all degrees-of-freedom (DOFs) to estimate structural parameters.However, in practice, it is often difficult, if not impossible, to make such measurements,due to sensing constraints, missing data problem, and/or excessive number of DOFslike in large-scale civil structures. This lack of the measurements in space results in illposedproblems with non-unique solutions. To address this challenge, this paperpresents a structural parameter estimation algorithm that incorporates spatiallyincomplete measurements and inaccurate prior information of the structuralparameters, within subspace system identification framework. Additional constraintsare imposed using prior information and the prior information is updated with the newestimation. This process is repeated as more measurements are collected, tosequentially update the parameters. The proposed method is evaluated using anumerical model of a five-story shear building for two damage scenarios withmeasurement noise. The structural parameters are estimated with 85~99% accuracywith spatially incomplete measurements (40~80%), and the iterative updating furtherimproves these accuracies.