Improvement of quality management in the cold rolling and finishing area by combining failure mode and effect analysis with data-based approaches

The main topic of this project was the combination of rule-based methods like failure mode and effect analysis (FMEA) or anticipatory failure determination (AFD) with data-based methods. The overall aim was to improve quality management in the cold rolling and finishing area and thus improve productivity and reduce production costs. A traditional FMEA was performed for tinplate and galvanising lines as well as for several stainless steel production plants. The resulting rule base was stored in a database. Furthermore, the AFD was applied for two different application examples by means of a commercial tool. The results of the FMEA and AFD were combined with the data-based methods like multi-variate statistics, artificial neural networks or decision trees, on the one hand, to verify the results of the FMEA and, on the other, to improve the quality of the FMEA by means of a comprehensible calculation of the risk priority numbers. Using the results of the investigations, a prototype application was developed and installed at several plants. With this the user can apply the database techniques, GUI-driven, without deeper knowledge of the parameters of the methods. The application was applied to solve actual quality problems in the production of flat steel. From the results of these investigations quality defects could be clearly reduced.