Sirovnik, E.; Brandenburger, J.; Rabe, M.: Considering Surface Inspection Results for Multi-objective Production Scheduling in the Dynamic Environment of the Flat Steel Industry, Conference presentation at Surface Inspection Summit SIS.EUROPE 2021, 9th-11th November 2021, Aachen, Germany.

The European steel industry represents a significant sector of the European economy. In order to maintain its competitiveness in the face of continuously increasing competitive pressure, it has to take on various challenges. The most important tasks in this context are minimizing costs and expanding unique selling points such as quality. Advanced production planning in flexible flow production can help to meet these requirements without the need to invest in capital-intensive equipment. The European Union (EU) funded Research Fund for Coal and Steel (RFCS) project DynReAct is developing a multi-objective scheduling approach to generate optimized production plans for each individual coil at each production step in a dynamic environment, considering real-time plant information. This approach is illustrated by the present case study of the product type Very Sensitive Products (VSP), which has very high surface quality requirements. One of the main reasons for quality problems with VSP are specific production-related spots on the surface of the material, which occur on the finishing lines. Based on historical and online surface inspection data, a model is developed to detect these specific spots early and provide an impulse for re-scheduling to reduce or avoid the associated quality problems. This plant performance model contains a multiscale data representation model and Deep Embedding Clustering (DEC) developed for industrial applications.