Eventos Anais de eventos
COBEM 2017
24th ABCM International Congress of Mechanical Engineering
Parameters identification with evaluation of uncertainties using sensitivity method and covariance matrix analysis
Submission Author:
Alexandre Löw , RS
Co-Authors:
Herbert Gomes, Enzo Costamilan, Alexandre Löw
Presenter: Alexandre Löw
doi://10.26678/ABCM.COBEM2017.COB17-0263
Abstract
An iterative sensitive-based procedure is presented in this work for the updating of some chosen input parameters of numerical or analytical models so that the predicted response approximates the measured data as closely as possible. At the same time, the uncertainty of these input parameters is estimated by matching, with the best approximation, the covariance matrix of the predicted output vector to that calculated for the measured data. The two main sources of uncertainties, random and epistemic, are incorporated to the numerical model using the elicited uncertain parameters and sensitivities, i.e., a jacobian matrix defined as the variation of the output parameters to some small variation of the inputs, approximated by finite differences. The main steps used in the processes of model updating using the sensitivity method for a simple dynamic model and experimental modal data are delineated, and then an application of this methodology to a three degrees-of-freedom model is presented for which experimental data was acquired. The updated model presented good agreement with the statistical data (mean value and covariance) obtained in the experimental test campaign
Keywords
Model Updating, uncertainty quantification, experimental modal analysis
