Application of Modal Scaling to the Pole Selection Phase of Parameter Estimation

Modern modal parameter estimation algorithms are frequently presented as two stage solution processes where the first stage identifies the system poles and unscaled modal vectors (participation factors) of either long or short dimension, and where the second stage identifies the scaled modal vectors (residue vectors) of generally long dimension and modal scaling. This paper explores the value of having the long dimension, scaled modal information available during the pole selection process. Among the advantages of this approach is the availability of the full length residue vector for visualization and the modal scaling in order to evaluate relative contribution and physical significance. A comparison of the residue quality for this solution approach and the dominant traditional approaches is presented. The methods and results are compared using mean phase (MP), mean phase deviation (MPD), and vector scatter plots.