Two-stage system identification approach for three-dimensional structural systems

Abstract

A two-stage finite element system identification (SI) technique is proposed in this paper to identify stiffness of elements and detect damages in three-dimensional framed structures. The technique combines in stage 1 the iterative least-square and in stage 2 the unscented Kalman filter (UKF) to identify the stiffness of elements using only limited measured response time histories from only four to six accelerometers instead of dozens of accelerometers of the whole structure and assuming the time history of dynamic load applied on structure is unknown. The method will identify the stiffness and detect the damages in the elements by tracking the changes in the recordable dynamic output responses between damaged and undamaged states. The optimum number and locations of accelerometers were studied in this paper. The algorithm is verified using numerical examples. The results showed clearly that the technique can identify damaged and undamaged three-dimensional steel framed structures and the minimum number of sensors required for such frames.