Structural damage identification via response reconstruction under unknown excitation

The restriction of limited sensors for measurement acquisition and the presence of unknown excitation usually bring about more challenges in the vibration-based damage identification of large-scale civil structures. To explore the potential benefits of utilizing multitype dynamic responses and alleviate the necessity of deploying dense sensors and measuring excitations for structural damage identification, this study investigates a novelty damage identification method via response reconstruction when the external excitations acting on the structure are unknown. Response and excitation are reconstructed simultaneously through the implementation of Kalman filter under unknown input. Radial-basis-function network is employed to predict the mode shapes using modal properties extracted by experimental modal analysis after damage occurrence. The reconstructed response and excitation are finally integrated into to sensitivity based finite element model updating for localizing and quantifying the damage. A numerical simulation study is conducted on an overhanging steel beam under unknown excitation. The feasibility and effectiveness of the proposed method are further ascertained by laboratory tests of the beam.