SimTech Posenidentifikation

Background

Machine tools have pose-dependent natural frequencies within their workspace, especially those with gantry or travelling column design. These can be explained by the cantilever structure and the changing inertia ratios. Variable natural frequencies are generally not considered during controller settings and the controllers are set according to the dynamically worst case.



Problem

In the future, not only the saving of energy but also resources will be major trend in the development of production plants. A mass reduction of structural components of machine tools, which exceeds the optimization of topology, results in decreased dynamic stiffness of the very same structures. This in turn has effects upon the occuring structural vibrations, which have to be countered with active vibration reduction methods. However, with many conventional vibration reduction methods, the current natural frequency and mode has to be known exactly. But these vary over the workspace of the machine and are still dependent on the machining state. The exclusive metrological detection of occuring natural frequencies is very time-consuming and expensive. The detailed modelling of the pose-dependent dynamical behavior of machine tools is also cost-intensive. Further is the conformity of model and reality in case of the complex system of a machine tool with multi-scale and multi-physics simulations rather questionable. Due to the computation time is the real time-capable integration of a very detailed model into industrial control technology currently not possible.



Results

This project combines the model-based with the experimental approach. The parameters of relatively simple, general machine tool models with gantry or travelling column design are adjusted to the current dynamic behavior of the machine tool by means of control-internal and additional signals. The models are realized as parametrical multi-body models with defined eigenmodes. Due to their design these eigenmodes can be predicted to be dominant. The model parameters are adapted online to the current real dynamic machine behavior through recursive least squares algorithms. The adapted model can in turn be used for real-time control or vibration reduction. The described approach is implemented at ISW at a laboratory prototype of a lightweight travelling column machine tool structure, which is shown in the picture.

 

The project is funded within the cluster Simulation Technology