Overview
This work analyses a fuse production line from the standpoint of statistical process control. Classical statistical control charts, as well as novel methods based on Kullback-Leibler divergence and anomaly detection are tested. To iterate fast and on the whole data that is available from the whole production line a SQLite database is employed. Unfortunately, none of the tested methods were able to achieve the goal of distinguishing a normal process from an unstable one. The causes for this are attributed to parameters not yet recorded, and further steps and recommendations are given.
