Predictive maintenance is typically associated with the analysis of large volumes of data and is therefore a prime example of big data. These data volumes need to be stored and evaluated over a long period of time. Cloud solutions, for which all well-known German manufacturers offer solutions, provide support for this computationally intensive task. A typical component for condition monitoring, which ultimately leads to predictive maintenance, is the ball screw drive.

Just last year, the VDW Research Institute, together with its members at the Institute of Production Engineering and Machine Tools (IFW) at Leibniz University Hannover, successfully completed the project "WiZuBe Economic Condition Monitoring of Ball Screws". Among other things, the results provide valuable insights into which characteristics are important for monitoring and how data volumes can be reduced through intelligent forgetting. In the coming years, the algorithms will certainly be improved and the predictions will become more accurate. Artificial intelligence will play its part in this.

Is machine learning for more precise forecasts a topic that machine tool manufacturers are also addressing?

Machine learning as a branch of artificial intelligence will improve forecasts. There is no doubt about that. Developments are underway and the pace is rapid. Machine tool manufacturers are not turning their backs on this complex of topics - on the contrary. The "IIP Ecosphere" research project is being actively shaped by the VDW and its members. This is about nothing less than the vision of implementing an innovative leap in the field of self-optimization of production through AI and spreading it throughout the industry, especially among SMEs. As part of this project, VDW member Gildemeister Drehmaschinen in Bielefeld is developing a demonstrator that combines feedback from experienced machine operators, process data and data from quality inspection, for example, as an AI-based assistance system. It should help to better adjust and monitor production processes.

Do machine manufacturers also offer options for retrofitting older systems?

There are options for retrofitting in the brownfield, but these must be assessed on an individual basis, as it is difficult to make general statements. In the past, external sensors often had to be retrofitted, for example temperature meters or vibration sensors. The "WiZuBe" research project has shown that external sensors can now often be dispensed with. The clever combination of signals available in the control system, such as motor currents and torques, makes external sensors obsolete. They no longer increase the reliability of the prediction. However, modern control systems for machine tools then make sense. However, these are usually replaced during a retrofit anyway.