Constrained unscented Kalman filter for parameter identification of structural systems
Dan Li, Yang Wang
Abstract
The unscented Kalman filter (UKF) can be used to identify model parameters of structural systems from the measurement data. However, the standard UKF may provide unreliable and nonphysical estimates, since no parameter constraints are incorporated in the identification process. This paper discusses and compares several constrained UKF (CUKF) methods for parameter identification of structural systems. The effectiveness and robustness of the methods are evaluated through numerical simulation on a Bouc–Wen hysteretic system. The results demonstrate that with properly handling of the constraints, the identification accuracy can be improved. The proposed CUKF method is further validated using experimental data collected from a full-scale reinforced concrete structure. Based on the identified model parameters, the updated models can achieve more accurate simulation responses than the initial model.