Machine learning based operator assistance in roll forming

Abstract This paper presents an approach for operator assistance in roll forming to overcome the challenges of progressive skilled labor shortage faced by manufacturers of profiled products. An introductory study proves the necessity and the willingness of the roll forming industry to use process data and machine learning based assistance for less experienced operators. A newly built framework contains the characterization of process behavior based on in-line collected data. To support operators during the setup and control of complex manufacturing processes, correlations between tool adjustments and process data are analyzed in a machine learning (ML) pipeline. Setup suggestions are directly provided to the operator for implementation and a feedback loop takes the results into account. To quantify the functionality of the newly developed Machine Learning based Operator Assistance (MLbOA), an exemplary roll forming process is investigated. The system localizes maladjustments in the setup of tool gaps caused by individual mechanical load behavior and offers corrective suggestions to operators with a mean absolute error of 1.26 ± 0.36 μm. This work demonstrates the potential of machine learning based assistance systems to enhance the resilience of manufacturing processes against the challenges posed by the shortage of skilled labor.