Optimal Data Selection for Piezoelectric Material Characterization

In the simulation and design of piezoelectric transducers the exact knowledge of the entries in the material tensors - the elastic, dielectric and piezoelectric coupling coefficients - is an important prerequisite. Our task is the identification of these coefficients from indirect measurements, namely electric impedance data at different frequencies. This leads to a parameter identification problem for a system of coupled PDEs, which we solve by regularized Newton iterations. A crucial issue is the selection of frequencies at which measurements are taken. Here, we discuss the problem of choosing these frequencies in an optimal way to preserve efficiency of our identification scheme while improving reliability of the reconstruction results. For this purpose, we formulate this task as an optimization problem with PDE constraints and propose two approaches for its solution. (© 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)