Non-Contact Based Structural Damage Detection Using Stochastic Subspace Identification and a FEM Updating Method

This paper proposed a method to identify dynamic characteristics and system matrices considering uncertainties of modal and ambient excitations. Three algorithms are integrated into this procedure: stochastic subspace identification (SSI), mass change modal scale (MCMS), and finite element model updating method (FEMU). Using only the output response of the structure measured under unknown ambient vibration, the SSI algorithm is used to identify the mode shapes, natural frequencies, and damping coefficient. In typical operational analyses, the mode shapes are not correct (i.e., unscaled mode shape). A mass change modal scaling algorithm is used to identify true mode shapes (i.e., scaled mode shape). With scaled mode shapes and frequencies, a FEMU method is used to determine the mass and stiffness matrices. By analyzing the changes of mode shape, coefficients on each node can detect and locate the damage. Numerical simulation shows that under different types of excitation and with white noise influence on the response, the damage detection method exhibits high sensitivity towards changes in mass and stiffness matrices. An experimental verification of this method is carried out with a small-scale cantilever beam. A high-speed camera is used to acquire the displacements of the elements. Finally, the proposed algorithm can be used to extract frequencies and scaled mode shapes in conjunction with displacements measured over time by passive non-contact optical methods through video.