Data-driven Grinding Burn Monitoring

 

The microstructural changes that occur in grinding technology are the result of thermal overload in the component edge zone and are referred to as grinding burn. The occurrence of grinding burn and the associated change in material properties lead to poor operating behavior of the components and thus to a reduced component service life. One of the challenges in grinding steel is therefore to avoid grinding burn while maintaining consistently high productivity.

Conventional inspection of workpieces for grinding burn is possible by methods such as nital etching and the measurement of the Barkhausen noise or X-ray diffraction. However, the established methods are either exclusively destructive or can only be automated and transferred to other processes with severe limitations. A comparatively cost-effective solution for process-independent grinding burn monitoring in series production is therefore not state-of-the-art due to the deficits of conventional measuring methods listed above.

Data-driven grinding burn identification offers the potential for process-independent inspection of workpieces for thermal edge zone damage. In particular, the acoustic emission signal can be used during the grinding process to detect damage at an early stage and avoid scrap. Implemented in an assistance system, the possibility is offered to monitor industrial series grinding processes during grinding.

The goal of the project is therefore to develop a method for data-driven grinding burn identification and to embed the method in an assistance system for data-driven grinding burn monitoring. Machine learning models are used to identify grinding burn with the help of acoustic emission.