Data-Driven Structural Deformation Identification for High-Rise Buildings Using Multisensor Measurements with Real-World Validation

Structural deformation of high-rise buildings is informative and crucial for assessing both safety and serviceability. This study introduces a data-driven two-stage estimation framework for flexural deformation of high-rise structures employing multisensor measurements. The first stage of the framework involves the identification of the quasi-static deformation curve at the system level using the inclinometer data and Euler-beam (EB) theory. In the subsequent stage, full-band lateral displacement at the story level is estimated by fusing dynamic acceleration data with the previously estimated quasi-static displacement through a multirate Kalman filter. The proposed method is numerically investigated on a 42-story benchmark model subjected to wind and earthquake excitations. The results demonstrate its capability to accurately identify quasi-static deformation curves at the system level and estimate full-band lateral displacement at the story level. Finally, the feasibility and effectiveness of the proposed framework are validated through field measurements from a 632-m-high skyscraper during Typhoon Muifa. The findings indicate that the proposed method accurately estimates the multilevel structural deformation, with displacement estimates comparable to global position system (GPS) measurements. Overall, the proposed method proves to be both feasible and reliable for monitoring structural deformation using a structural health monitoring system, thereby providing valuable insights into the structural health conditions.