Construction of frequency response function confidence interval by a semi analytical approach
Résumé
This study proposes an original approach for the construction of transfer function confidence interval. Uncertainty quantification in modal superposition involves two steps. The first one is the resolution of a random eigenvalues problem. The second one is the calculation of the probability distribution of the transfer function thanks to the results of the random eigenvalues problem. We propose to focus on the second step and to solve a simplify version of this problem. This simplified study, allows to obtain several conclusions which can be extended to the actual problem and give important informations to lead to the resolution of the random eigenvalues problem. In particular, the influence of random eigenvalues marginals is quantified and the importance of the dependence structure between eigenvalues is highlighted. Moreover, an analytical criterion is set up to define the range of influence of each random mode. Finally, the methodology is tested for different applications and the accuracy of the results is quantified by bootstrap confidence interval.