Predicitve Maintenance of Machine Tools using a digital twin

Description

Big data approaches have not yet led to a breakthrough in predictive maintenance. The remedy is the holistic, physically-motivated consideration of the feed drive with the aid of a virtual representation. A special test cycle ensures that the measurement of disturbing influences is reduced to a minimum and that actual wear can be measured. Wear-sensitive "features" are generated from the measurement data. In contrast to data-driven analyses, their design is based on system understanding and a digital twin of the machine tool. Both the time domain and the frequency domain are considered. A final evaluation of these "features" allows the determination of the condition of the feed axis and an estimation of the remaining useful life of its components. The digital twin also allows conclusions to be drawn about the manufacturing accuracy of the machine still to be achieved.