Data-Based Model Parametrization of Quality Predictive Material Removal Simulations

The ability to control product quality in milling processes is an essential target variable. Quality deviations can have different causes, which can be both systematic and stochastic. Tracing data from numerous sensors of modern machine controls and combining them with a material removal simulation (MRS) allows to predict the workpiece quality in a process-parallel manner. The accuracy of the prediction depends on the underlying model and its parametrization. Model parametrization usually requires measurements during which the machine tool cannot be used. Additionally, the necessary measuring equipment is often not available in companies. All this makes the implementation of MRS difficult in the manufacturing industry. This article presents an approach that circumvents the abovementioned disadvantages. With the help of a MRS, machine data is refined and contextualized with quality data. The refined and contextualized data is then used to fit various models, such as stiffnesses and process force. In future the presented approach should enable a cost-effective parametrization without machine downtime. With the parameterized models, MRS that predict quality can be used both in advance during CAM planning and in a process-parallel manner for quality monitoring.