This paper presents a dispersion-based method for identifying visco-elastic material properties by minimizing the residue between data of virtual experiments and data based on a viscoelastic model through the use of surrogate modeling. The dispersion data retrieved from a virtual experiment of a finite beam with constrained layer damping (the real and imaginary part of the wavenumber) is fitted with numerical dispersion data through an optimization scheme, which can be computationally expensive. In order to alleviate this issue, attention has been focused on the construction of a surrogate model that makes the optimization scheme cheaper without loosing much accuracy in the prediction. This paper uses an interpolation method based on radial basis functions. Once the surrogate model is constructed, the viscoelastic parameters can then be identified and results are compared to the reference parameters.
展开▼
