Development of Decision Support System based on Machine Learning and Digital Twin for aluminium melting furnaces
The sustainability vision that gained importance with the European Green Deal, has also affected the Aluminum sector, which is considered as one of the most energy intensive industries in Europe. Starting from this vision, the European Union called for projects, leveraging the Horizon 2020 program, with the concept of using existing raw materials more efficiently and enabling the use of alternative raw materials within various industrial sectors. RETROFEED is one of the projects supported by European commission under Horizon 2020 program. Among the solutions under development for the aluminum sector there are: (1) the usage of alternative raw materials in ASAS' aluminum melting furnaces, (2) equipment retrofitting allowing the existing resources to be processed more energy-efficiently, and (3) the design of decision support strategies in order to use the existing raw materials for producing less waste material during production. Strategical and operational decisions are made by the factory personnel, at times causing slowdown in production and/or inefficient use of resources. Within the scope of the project, a Decision Support System (DSS) will be developed integrating machine learning methods, for predicting billet quality according to different raw material types used in ASAŞ Aluminum Melting Furnaces, and a furnace Digital Twin to simulate furnace operations under different conditions for efficiency improvements and production simulation. By means of correlations and algorithms established as a result of the study, the output of the billet chemical composition, the furnace setpoints and the production plan can be adjusted according to different inputs in 6060 aluminum alloy, in accordance with the principles of zero waste and the new green deal guidelines.