Identificação não-paramétrica de sistemas mecânicos usando filtros de Kautz

Abstract

Impulse Response Functions (IRFs) are important in many engineering applications, mainly in structural dynamics and modal analysis involving experimental modal tests. These IRFs can be identified through several methods. Among these, the classical covariance method is one of the most used and it is based on the sum of convolution from the correlation functions between input and output signals known. However, this method is limited because it employs a large number of samples and has drawbacks related to over parametrization. In this sense, this work presentes and review the covariance method expanded in the ortonormal basis Kautz functions, because this alternative way allows to avoid these drawbacks. In order to ilustrate the procedure an algorithm with multiple objective functions to obtain the optimal poles of the Kautz filter is shown. The results are provided through three degree-of-freedom mechanical system simulated and experimental data in a beam to show the advantages, drawbacks, simplicity and efficiency of the proposed approach.