Eventos Anais de eventos
DINAME 2023
XIX International Symposium on Dynamic Problems of Mechanics
Parameter Estimation in Bolted Joints by Bayesian Inference
Submission Author:
Vitória Carolina Duarte , DF
Co-Authors:
Vitória Carolina Duarte, Marcela Rodrigues
Presenter: Vitória Carolina Duarte
doi://10.26678/ABCM.DINAME2023.DIN2023-0150
Abstract
The tools to describe objects range from analytical solutions for simple studies to computational tools to model the most complex systems. Physical data is used to calibrate numerical models that can use a variety of forward or reverse bias methods. In mechanical vibrations, one application of the inverse problem is when data such as displacement or frequency spectrum are obtained using accurate sensors with known application domain, the complementary conditions are known - controlled experiment - and the desired properties are desired. e.g. equivalent stiffness, viscous dissipation coefficient, natural frequency. In more complex systems such as bolted joints, it is clear that approximate solutions do not always apply. Thus, the experimental case of a dynamic mechanical system was studied to obtain the estimated stiffness values (kv and kt) by Bayesian inference via MCMC - MH, where the input data were obtained experimentally, with the torque of 80Nm in the bolted structure, so the solution to the uncertainty quantification problem can be assisted by the UQLab toolbox.
Keywords
Inference Bayesian, Inverse problem, Bolted joints, MCMC
