Parameter Estimation of Chain-Like Structures Based on Incomplete Measurements Using Subspace System Identification

Abstract Infrastructure systems such as bridges and high-rise buildings are susceptible to structural damage over long periods of service. To ensure their safe operation, it is essential to continuously monitor structural dynamics and detect potential damage based on changes in dynamic behavior. Detailed information about structural damage can be obtained from parameters including mass and stiffness, which are extracted from the identification results. This process of parameter estimation often requires measuring the responses of all degrees-of-freedom (DOFs). However, installing sensors at every DOF in large-scale structures is challenging. Moreover, conventional methods may fail to yield a unique solution when the number of sensors is insufficient. To address this issue, we propose a parameter estimation method that requires fewer sensors than the number of DOFs. The target structure in this study is a chain-like structure, a commonly used modeling approach for real-world systems such as high-rise buildings and offshore platforms. The method incorporates new prior information derived from physical laws and consists of two key steps: an algebraic approach followed by an optimization approach. In the algebraic approach, a general solution containing undetermined variables is obtained owing to the limited sensor configuration. Subsequently, these variables are solved in the optimization step using an objective function constructed from prior information. The effectiveness of the proposed method was validated through both simulations and experiments using only a single sensor.