Adaptable tool monitoring systems for turning processes based on inprocess cutting force sensors

• Applicants: Professor Dr.-Ing. Thomas Bergs; Professor Dr.-Ing. Welf-Guntram Drossel
• Subject Area: Metal-Cutting and Abrasive Manufacturing Engineering ; Production Automation and Assembly Technology
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 524221060

The overall objective of this project is to explore fundamental insights to increase the generalisati-on capability of sensor-based AI-supported tool monitoring in turning processes. The focus is on an adaptive monitoring system with which offline trained decision-making models for predicting tool life can adapt in the online process and react to new cutting parameters to a certain extent. State-of-the-art methods of "artificial intelligence" (AI) are ideally suited to increase the generalisation capability of tool monitoring systems. For the successful development of AI-supported systems for this purpose, it is essential that the data is available in sufficient volume as well as the necessary quality. This is often not guaranteed within single-part and small-batch production. The use of more process-sensitive and compact sensors for cutting force measurement in the immediate vicinity of the tool cutting edge provides the basis for an increase in data quality and a reduction in the amount of data required for the training of AI-supported process models. A second pillar of signifi-cant importance for increasing the generalisability of the AI approaches is to utilise the potential of a continuously learning system, e.g. for unlearned new training examples with new cutting parame-ters in the real production environment. The automation strategies available at the WZL (RWTH Aachen) for data acquisition from tool life tests and the competences for AI-based process mo-delling for monitoring the tool wear condition serve as important pillars for achieving these goals. Here, the relationship between data quality and quantity also plays an essential role in the model quality of the decision-making methods. In order to optimise this conflict of objectives, the sensor approaches for near-impact cutting force measurement at the Fraunhofer IWU again offer a promi-sing paradigm for increasing the quality of the measured data. The cooperation of both research institutes to investigate the relationship between data quality, effectiveness of the automated data acquisition strategies and performance of the process models is ideally suited with regard to an adaptable online tool monitoring system.

• DFG Programme: Research Grants