OPERATIONAL MODAL ANALYSIS UPON ANALYTIC SECOND-ORDER BLIND IDENTIFICATION AND DISCRIMINATION BETWEEN PHYSICAL AND SPURIOUS MODES

The automated extraction of operational modal parameters is of great importance to real-time health monitoring of large infrastructures. Regarding to the shortage of the existing methods in view of the difficulty of modes discrimination, an operational modal analysis method based on analytic second-order blind identification is proposed. A companion method to discriminate between physical and spurious modes for automated identification is also developed. Unlike traditional methods, the modal responses of each mode can be separated directly from structural vibration responses without using stabilization diagrams. Physical and spurious modes can be discriminated by introducing modal metrics and adopting K-means clustering. Modal parameters can then be estimated by the frequency domain parameter fitting method. The proposed method is validated by a numerical example of 8-DOF system and an engineering example of concrete arch-gravity dam. The results show that the method has preferable ability in automated modes separation and discrimination and, demonstrates high precision in the identification of natural frequencies, of damping ratios and, of mode shapes, showing remarkable advantages, especially in the appearance of closely spaced modes.