Identification of Mechanical Systems through Kautz Filter

Abstract

This paper presents a novelty approach to estimate the impulse response function (IRF) in lightly damped structures by using classical covariance method. The contribution of the present paper is to employ a Kautz filter to reduce the number of terms in the identification process of a finite impulse response (FIR) model. The Kautz filter is an orthogonal function that includes complex pole pair. So, the method incorporates a priori knowledge into the process that is essential to reach better results and to minimize the number of coefficients in the FIR model. For this reason, the present approach can be classified as a kind of "grey-box" methodology. The numerical and experimental applications of the method confirmed that by considering a priori parametric information the problem can be determined more easily. Moreover, the procedure proposed increased the convergence speed, and, as a consequence the consuming computational cost is lower.