Abstract For production of sheet metal parts for car bodies, an adjustment of process parameters is required to maintain the desired part quality in presence of scattering blank properties. The digital transformation enables the application of data-driven methods for finding process parameters instead of a time-consuming experience-driven trial-and-error approach. However, due to cost and technical limitations, it is still hard to measure quality for every part. Removing data points of low-quality parts helps recommending proper process parameters. In this paper, we propose classification-based solution for recommending process parameters. In data preprocessing, the solution utilizes anomaly detection and knowledge-based methods to remove potential data points of low-quality parts without quality measures. On the processed data, a classification model is trained to predict process parameters according to blank properties. Our solution detects 30% low-quality parts and gives competitive performance (92.26% prediction accuracy) compared to a model trained on data comprising quality measures.
Read full abstract