Optimal Sensor Deployment: Application to Model Updating

Abstract

The proposed formulations deal with various specific aspects of the problem of parametric identi fication of linear elastodynamic finite element models. They involve adapting, via expansion data observed on the structure, to the analogous quantities calculated from the model. They also aim at improving the condition number of the estimation equations. The minimization of expansion errors in frequency responses or eigenvectors by an optimal choice of the observed dofs is formulated in two cases: the first uses a Ritz basis consisting of eigenvectors and static residuals, while the second implies using a Guyan-static condensation matrix. The relationship between optimal expansion and maximization of the cutoff frequency corresponding to this condensation is also analyzed.