ABSTRACT
ABSTRACT This paper presents a novel filtering algorithm for joint input-state-parameter estimation. The algorithm is derived from an existing joint input-state estimation algorithm. In each step, the system model is linearized around the current state, yielding an algorithm which is similar to the Extended Kalman filter. It is shown that, in the application to linear structural dynamic systems, the analytical expressions for the Jacobian matrices involved in the linearization are readily available due to the choice of parametrization. The proposed methodology is verified by numerical simulations for a four story shear building.
