For the condition assessment of existing infrastructure buildings the current parameters of the building material, such as the Young's modulus and Poisson's ratio, are of great interest. These parameters indicate damage or material degradation, since they reflect the resistance of the structures to external impacts. Provided the displacement field data is available, the material parameters can be identified by solving the momentum equation inversely. It was recently shown that there is a method in the growing field of scientific machine learning, known as physics-informed neural networks [1], that is particularly suitable for the inverse solution of partial differential equations. In contrast to purely data--driven approaches, physics-informed neural networks do not only use the displacement data but also the physics behind the data formulated as partial differential equation. It has been shown that physics-informed neural networks can in principle be used to identify material parameters [2], [3]. In the present paper, we first develop a three-stage adaptation of the method to realistic one-dimensional applications. Second, we verify the adapted method on two-dimensional displacement data. We conclude the paper with an outlook on ongoing and future work to further improve the method.