At the NAMUR Hauptsitzung, which took place on 23-24 November 2023 in Neuss, Kai Dadhe (Evonik) and Sebastian Engell (TU Dortmund DYN) gave a talk on the results of the KEEN project, which recently ended in September 2023. The presentation entitled "Is AI now controlling our plants? - Lessons learnt from the KEEN Project" provided insights into selected KEEN developments and case studiess in the areas of AI or machine learning (ML) based data analysis and control, and summarised important results and findings from the project. The presented examples illustrated that the application of AI/ML methods has the potential to speed up the classification of data and can lead to better models for control faster. However, trust in models plays a fundamental role, especially for online applications. Sebastian Engell emphasised: "Data-based models can only predict what they have seen!" One of the proposed solutions that was investigated during the KEEN project is the use of transfer learning methods, in which a machine learning model pretrained with simulation data and then re-trained with data from the real plant. While the realizationof self-optimisinging plants is still a long way to go, AI and ML supported plant operations promise important improvements that can be achieved with lower efforts than needed for rigogorus modelling, if sufficiently rich data is available.
Presentation: "Is AI now controlling our plants? - Lessons learnt from the KEEN Project"