Identification of very closely spaced modes using an iterative algorithm

Abstract

The algorithm of mode isolation is a frequency domain method for processing measured response data in order to identify the modal properties of a system. It has been shown [M. V. Drexel and J. H. Ginsberg, Proceedings of the 19th IMAC, Orlando, FL, 5–8 February 2001] to accurately evaluate a pair of damped modes whose natural frequencies differ by an amount that is commensurate with the bandwidth of either mode. Here, SIMO measurement of a two-degree-of-freedom system is simulated analytically by adding white noise to the computed response. This system is selected because the natural frequency difference can be made as small as desired by adjusting a parameter that has little effect on the modal damping ratios in the range of interest. Analytically, the two normal modes are unique and mutually orthogonal if the frequency difference is nonzero, while repeated frequencies lead to two arbitrary, and not necessarily mutually orthogonal, modes. The present work explores the degree to which AMI tracks the analytical behavior, specifically, its ability to detect and characterize the modes with decreasing frequency difference.