Condition Based Maintenance for Industrial Labeling Machine

This paper reports the outcome of an industrial research on data-driven Condition Based Maintenance (CBM) for the film cutting group of labeling production lines. Objective of the study has been the prediction of erroneous labels cut. The large number of variables involved in thin labels cut (thickness comprised within 30μm and 38 μm) and the high throughput make the prediction of non conforming labels a difficult goal. To this aim, we developed a complete CBM strategy for film cutting groups. To identify failure signature, an exhaustive assessment on indices suggested in literature was done, but none of them were suitable to satisfy problem constraints. Thus we customized the most promising one (namely the root mean square value of the vibration measures) to our setting obtaining notable results. Given the lack of contributions in CBM in thin film cutting, we believe this paper might be of interest for academic researchers or people from industry dealing with similar problems.