Non intrusive uncertainty quantification method for models with a high number of parameters - Application to a magnetoelectric sensor

摘要

To face the "curse of dimensionnality" met in uncertainty quantification problems when the model has a high number of random parameters, methods based on sparse approximation, like the Least Angle Regression (LAR) method, should be used. In this communication, we propose to extend the domain of applications of such methods and to apply them to quantify the impact of uncertainty on a magnetoelectric sensor performances. The sensor response is represented by a 2D finite element model with 10 random parameters. A global sensitivity analysis is carried out in order to determine the most influential parameters.