Bias in modal parameters using the direct and iterative RFP identification procedures

In the rational fraction polynomial method [1] the identification of modal parameters is obtained through a direct linear least-square optimization technique but a particular form of fitting error is minimized. An iterative algorithm has been recently developed which uses the true fitting error [2]. In this paper a statistical analysis is developed to estimate the bias effects on the identified parameters when the data are polluted with noise. Both the direct and iterative procedures are considered. Numerical simulations are used to validate the results predicted by the theoretical analysis, which shows that the iterative approach is by far more efficient than the direct method. © 1995 Kluwer Academic Publishers.