Generating parametric models from tabulated frequency domain data

Fraunhofer ITWM, Kaiserslautern, Germany

Creating parametric models which incorporate the dependence on parameters is of great importance during the design phase of a product, when certain variables (eg. sound pressure level inside a car at the location of the passenger’s ears) needs to be optimized. Design parameters could relate to the geometry of the product (length, width of the piezoelectric patch) or deal with material properties (Young’s modulus, Poisson’s ratio). Models for many configurations of the design parameters are needed inside an optimization loop, so making measurements to extract the necessary model for each of these parameter configuration is not feasible since it will take a long time. Thus, having parametric models available ensures a faster design workflow. After describing our approach, we will investigate the preservation of properties of the original system (mainly stability). Moreover, we will show how doing some steps differently gives better or worse results on some numerical test cases which we analyzed.

Laboratoire de Mécanique des Structures et des Systèmes Couplés - LMSSC