Design of a semi-physical dynamic model for a pneumatic bulge test using local model networks trained with FEM data

The modelling of hydraulic and pneumatic systems is essential for the machine development and control design process. In most cases, lumped parameter models provide sufficient accuracy. However, for some fluid power systems this approach doesn’t result in a useful model for example due to non-linearities of the system. An approach to overcome this issue is the use of local model networks. In this paper, a brief introduction into local model networks is given presenting the basic idea and the approximation algorithm used to derive the network. In the following, a calculation scheme is presented that allows fast evaluation of the network and thus enables the implementation into machine control for small cycle-times. The local model network approach is used to model the pneumatic hot gas bulge test, which is used in material characterisation. A database generated from FEM simulation is applied to set up the local model network. The final result is the comparison and discussion of the local model network output and a training set generated from FEM simulation.