Model updating for hybrid testing with unscented Kalman filter
WANG Tao1,2,WU Bin1
1. School of Civil Engineering, Harbin Institute of Technology, Harbin 150090, China;2. School of Architecture and Civil Engineering, Heilongjiang Institute of Science and Technology, Harbin 150027, China
Abstract:In a hybrid simulation, the whole system is divided into physical parts and numerical parts. For hybrid simulation of large-scale structures subjected to strong earthquake, it is usually difficult to keep the numerical substructure within linear range. So the concerns are raised how the modeling of the numerical part is reliable. A model updating for hybrid testing method with the unscented Kalman filter (UKF) is presented in order to improve the accuracy of model for numerical substructures. The new testing method assumes that the numerical substructures have the similar hysteretic behavior with the physical substructures, but the parameters of the hysteretic model are updated after they are identified with the unscented Kalman filter approach based on the data from experimental parts. The performance of UKF is investigated by numerical simulation for a single freedom nonlinear system. The effectiveness of the proposed test method is verified by an actual testing. The results show the method has good accuracy compared to conventional hybrid testing.