Crack detection in existing structures using noise-contaminated dynamic responses

Crack detection is existing structures using measured dynamic response information is presented in this paper. The general area of health assessment of existing structures in the presence of uncertainty in modeling the structure, characteristics of sensors, and the dynamic response information collected by the sensors is emphasized. A time domain system identification-based procedure is proposed to assess health of a structure at the finite element level in the presence of all major sources of uncertainty. It is denoted as the GILS-EKF-UI method. The method does not require information on excitation that caused the response and the noise-contaminated response information needs not be available at all dynamic degrees of freedom. The method is verified using computer generated analytical and actual measured response information emphasizing three items: (1) identification of the defect-free frame, (2) detection of location of a crack accurately within a defective element, and (3) detection of a crack using limited response information using the GILS-EKF-UI method.