Eigensystem realization algorithm applied directly to time-domain data using simulated inputs

Abstract

The modal parameters of a simulated four degrees of freedom dynamic system are identified by ERA algorithm using Markov Parameters obtained straightly from the time data. Random forces excite this simulated system. A method for evaluatiation the model order and for obtaining optirnized size of the Hankel matrix was employed to improve the accuracy and the efficiency of modal identification when the inputs are taken as white noise. To study the problemof modal parameters extraction in structures under operational conditions, a identification process was carried out using inputs lightly different than those applied to the system. For this it was necessary the application of Wavelet Transforrn (WT) for compression of the inforrnation making possible a complete identification including the low frequencies. Also this transforrn facilitates the application of a statistical analysis for selection of the best system order and for process automation. The results obtained were validated by comparison with the theoretical parameters of the system.