Accuracy of multiparameter eigenvalues used for dynamic model updating with measured natural frequencies only

Analytical models of linear elastomechanical systems can be updated by model parameter estimation provided that the structure of the model equations and the parametrisation of the spatially discretised model are consistent with the system under test. Due to the difference between the number of degrees of freedom (dof) of the analytical model and the number of measured dof there is usually a lack of measurement information not only for large-scale models but also for models e.g. finite-element-models (FE-models) with not measured rotational dof. However, only a minimum set of test data is needed for estimating the actual model parameter values if the original inverse problem of updating the linear analytical model is transformed to a corresponding multiparameter eigenvalue problem.In this paper, it is discussed how many natural frequencies of a system have to be measured and which accuracy of measurement is necessary to obtain reliable model parameter estimates for a given set of model parameters.