System Identification and Experimental Modal Analysis of a Frame Structure

摘要

In this paper, an effective method for performingthe experimental modal analysis of structural systems is developed. The proposed methodology is corroborated by analyticalconsiderations and is verified experimentally. The identificationmethod developed in this paper is based on time-domain systemidentification numerical techniques. The case study consideredin this work is a frame structure that can be modeled as a twostory shear building system. A preliminary mechanical modelof the two-story shear building system is developed by usinga lumped parameter approach. Subsequently, a more realisticsecond-order model of the frame structure is obtained directlyfrom input-output experimental data. To this end, a numericalprocedure based on the combination of the Observer/KalmanFilter Identification Method (OKID) with the EigensystemRealization Algorithm (ERA) is employed for determining thesequence of system Markov parameters. In particular, thefundamental matrices that characterize the state-space representation of a general linear time-invariant dynamical systemare obtained from the identified system Markov parameters.In addition to the identified first-order state space model, asecond-order mechanical model of the frame structure is experimentally obtained employing a methodology for constructingmechanical models from identified state-space representations.More importantly, considering the assumption of proportionaldamping, an effective method based on a simple least-squareapproach is used for calculating an improved estimation ofthe identified damping coefficients. The experimental modalparameters found by using the proposed methodology areconsistent with those predicted by using the analytical approachbased on the simplified lumped parameter model. Furthermore, the mechanical model identified employing the approachdiscussed in this paper is used for developing an activelycontrolled inertial-based vibration absorber based on the LinearQuadratic Gaussian (LQG) control and estimation method. Thenumerical and experimental results found in this investigationconfirmed the effectiveness of the methodology developed in thepaper.