An automatic frequency domain modal parameter estimation algorithm

In this article, a frequency-domain modal parameter estimation method is proposed. The algorithm automatically separates physical poles from mathematical ones. An important issue in the automatization of the algorithm is the inclusion of noise information to estimate the standard deviations of the poles. These standard deviations are used (together with other features) as the inputs of a fuzzy clustering algorithm. The clustering algorithm then classifies the poles into the mathematical and physical ones. The method requires no user interaction, and a parameter is available quantifying the success of the classification.