Experimental identification of dynamical structures in presence of uncertainties related to the placement of sensors and actuators

Anas Batou¹

¹ Université Paris-Est

doi:10.20906/CPS/USM-2016-0001

Resumo

The objective of this research is to identify the parameters of a dynamical computational model in presence of uncertainties related to the position and the orientation of the sensors and actuators. Such uncertainties yield uncertainties in the correspondence between the experimental dynamical responses and the dynamical responses calculated using the computational model. These uncertainties in the response increase with frequency and have to be taken into account when updating the parameters of the computational model in order to obtain a robust estimation of these parameters. To take into account and analyze such uncertainties, a probabilistic model is constructed directly on the reduced-order model yielding two localization random matrices. This methodology is little intrusive with respect to the commercial softwares and is well adapted to parallel computing. Furthermore, an optimal sensor placement method is proposed so that (1) the measured data are as sensitive as possible with respect to updating parameters and (2) the measured data are as robust as possible with respect to position/orientation uncertainties. The methodology we have proposed which is based on the use of Kullback-Leibler divergence, allows to rank a set of locations for the sensors and actuators. The methodologies developed here are illustrated through numerical applications.

Palavras-chave: Uncertain sensor position; Bayesian experimental design; Structural dynamics