RECENT IMPROVEMENTS OF THE ALGORITHM OF MODE ISOLATION

摘要

The Algorithm of Mode Isolation (AMI) identifies the natural frequencies, modal damping ratios, and mode vectors of a system by processing complex frequency response data. It uses an iterative procedure based on the fact that a general frequency response function is a superposition of modal contributions. The iterations focus successively on a single mode. The mode that is in focus is isolated by subtracting the other modal contributions using prior estimates of their modal properties. This process leads to a self-contained identification of the number of modes that participate in any frequency band, whereas other techniques require a priori guesses. This paper describes modifications intended to improve AMI's accuracy and reduce its computational effort. These involve the use of a new linear least squares method for identifying the natural frequency and damping ratio of a single mode, and a linear least squares global fit of the data in order to identify mode vectors. Results are presented for a model of a cantilever beam with suspended spring-mass-dashpot systems. This system was used by Drexel, Ginsberg, and Zaki [Journal of Vibration and Acoustics, 2003 (forthcoming)] to assess the prior version's ability to identify weakly excited modes and modes having close natural frequencies in the presence of high noise levels. Application of the modified version of AMI to the same system is shown to lead to significantly more accurate damping ratios and mode vectors, with equally good natural frequencies.