Extended Kalman filter（EKF） theory is developed from linear Kalman filter theory. When the physical parameters of the structure are unknown， the unknown parameters of the structure are combined with the state vector of the structure to form an augmented state vector， which is called extended Kalman filter theory. The unknown structural parameters are obtained through the augmented state vector. This paper introduces EKF theory to identify structural parameters. In order to verify the effectiveness of this method， a three-degree of freedom system with time-varying parameters is introduced as a simulation example to investigate the influence of various parameters （covariance matrix Q of model noise， covariance matrix R of measurement noise， influence of observation points） on tracking performance of time-varying parameters of structure. The results show that the algorithm with appropriate filter parameters （Q， R） can identify the time-varying parameters in the structure more quickly and accurately. And a cantilever beam simulation example is also introduced to show that the EKF algorithm can identify structural parameters even if some measuring points are not measured.