Prediction of damages on feed axes drives based on historical and synthetic data

The constant availability of production systems and machine tools ensures manufacturing without disruptions and with increased resiliency. To ensure the availability of critical machine components, it is important not only to detect damages but to predict them before they occur. Advancements in artificial intelligence, particularly concerning time series forecasting models, enable a precise prediction of damages, highlighting significant potential for their application in predictive maintenance. This paper presents an approach for forecasting damages in feed axes drives, which integrates the utilization of synthetic data. Historical data serves as a foundation for training the predictive models, while also facilitating the validation of the results generated by the algorithms. This dual reliance on historical and synthetic data not only enhances the robustness of the predictions but also addresses potential limitations associated with the availability of real-world damage data.