Eventos Anais de eventos
COBEM 2017
24th ABCM International Congress of Mechanical Engineering
Uncertainty compensation in damage identification by means of the approximation error approach
Submission Author:
Gabriel Lucas Sousa da Silva , RJ
Co-Authors:
DANIEL CASTELLO, Lavinia Maria Alves Borges, Gabriel Lucas Sousa da Silva
Presenter: Gabriel Lucas Sousa da Silva
doi://10.26678/ABCM.COBEM2017.COB17-0363
Abstract
This work presents an application of the approximation error approach (APA) proposed by Kaipio and Somersalo [1] in the context of Structural Health and Monitoring (SHM). Based on the Bayesian framework of statistical inversion, this approach allows one to compensate for errors caused by incorrect modelling of a physical system while still providing a relatively simple mathematical formulation. The application of different prior distributions of the unknown parameters is investigated with the help of a toy problem. The APA is compared to a traditional least-squares approach consisting of a forward model (FM) unable to compensate for model related errors. To reduce the dimension of the parameter space, a Gaussian bell curve was used to model the damage field in the FM.
Keywords
SHM, Approximation Error Model, Damage Identification
