Estimating Earthquake-Induced Displacement Responses of Building Structures Using Time-Varying Model and Limited Acceleration Data

Abstract

This paper presents a time-varying model-based method for estimating earthquake-induced displacement responses of building structures using limited acceleration data. First, a time-varying model which considers the variability of stiffness by a time-variant story local stiffness reduction factor and the variability of damping by formulating the Rayleigh damping model with a time-variant mass coefficient was developed. Then, a displacement estimation algorithm was derived on the basis of the state space representation of the time-varying model and the unscented Kalman filter. The effectiveness of the proposed method is numerically verified through nonlinear responses of a ten-story building model where the hysteretic restoring force is simulated by the Bouc–Wen model. In addition, numerical results indicated that a few sensors with low noise are more desirable for the measurement strategy. Finally, the applicability to realistic seismic monitoring of buildings was investigated through a 1∶3-scale 18-story steel frame test at the E-Defense shaking table facility, which tested collapse behavior and realistic damage of high-rise steel buildings constructed in the 1990s in Japan.