User-Specific Parameterization of Process Monitoring Systems

Abstract Errors in milling processes such as tool breakage or material inhomogeneities are a major risk to the quality of machined workpieces. like broken may also lead damages machine tool. Process monitoring systems allow for autonomous detection errors, therefore, promoting production. The parameterization these is trade-off between high robustness (low false alarm rate) and sensitivity. Even though several have been introduced single-item series production, universal technique that weighs off sensitivity does not exist. In this paper, novel, model-independent adjustable introduced. basis removal rate indicates temporal quantitative impact process errors (ground truth). ground truth allows calculation established F β -score, which used evaluate system. An adjustment β-parameter influences weighting robustness. Accordingly, easily control sensitivity-robustness so system economic company’s specific situation. look-up table hyper-parameters state-of-the-art tolerance range model provided using approach. With companies researchers can set their models 5-axis-milled single items user-specifically. To demonstrate, approach works different kinds models, one-class support vector (SVM) parameterized also.