Use of cross-time-frequency estimators for structural identification in non-stationary conditions and under unknown excitation

This paper introduces a new structural identification method for use in non-stationary conditions, which applies to structures and systems in normal serviceability conditions, under unknown excitation. The proposed method uses auto- and cross-time-frequency transforms of accelerometer signals recorded from the structure to identify the vibration modes. The transforms considered are those in Cohen's class which, in addition to possessing valuable properties for the analysis of mechanical signals, lend themselves to a clear interpretation in energy terms. This method enables modal parameters to be reliably estimated and an earlier technique proposed by the authors which was based on modal filters to be improved. It is further shown that the cross-correlation-based estimators are more effective than techniques based on auto-transformations, due to their noise-filtering properties. Finally, a method to refine the estimate of modal shapes, which avoids autocorrelation, is proposed. The accuracy of the procedure was assessed by means of numerical simulations. (C) 2000 Academic Press.