Noncontact Measurement Method for Inverting Structural Base Shear

Abstract

In response to the intricate installation challenges and the elevated cost of sensors for measuring base shear in large‐scale structures, this paper proposes a noncontact measurement method integrating computer vision and model updating to invert structural base shear. The computer vision part measures physical displacement, while the nonlinear model updating section inverts base shear by refining the structural numerical model, thus achieving cost‐effective, noncontact inverting measurements. In the computer vision component, a highly real‐time and accurate optical flow estimation algorithm was selected and validated in actuator motion tracking tests, yielding a normalized root mean square error of less than 3% between displacement tracking and sensor measurable results. The model‐updating section adopts the Bouc–Wen model, demonstrating through numerical simulations its ability to swiftly calibrate the numerical model within 7000 steps under various noise interference levels, accurately obtaining structural base shear. Moreover, the influence of different response combinations and sampling frequencies on parameter identification for model updating is discussed. Findings indicate that when considering both displacement and acceleration, along with a sampling frequency of 200 Hz, parameter identification meets accuracy requirements due to reduced susceptibility to measurement noise. In addition, a shake table test on a three‐layer shear frame is conducted to further validate the proposed method’s feasibility. Test results demonstrate that the amplitude and fluctuation trend of the shake table test’s identification results mirror those of the numerical simulation results within the first 25 seconds, with a peak value error of 18.9%. While the error is relatively large, this paper provides a practical research framework for model updating and structural health monitoring. Simultaneously, it reduces the cost of acquiring structural response data during tests, thereby facilitating the application and promotion of computer vision technology in structural response monitoring.