Unternehmen&Trends Digitalausgabe 01/2022

36 AI in production The next stage of AI-integrated manufacturing: Improvement of established processes, rapid industrialisation of “immature” processes By Dr. Olaf Sauer, Fraunhofer Institute of Optronics, System Technologies and Image Exploitation (IOSB), Karlsruhe Many companies are enthusiastic about the possibilities of artificial intelligence (AI), but are reluctant to promote its widespread implementation. Too often, initiatives which are started (still) remain at the ‘Proof of Concept’ stage instead of being widely utilised. In Karlsruhe, “AI System Engineering” is being driven forward as a discipline for effective use of AI in the manufacturing industry, e.g. for rapid optimisation or fast ramp-up of production processes. The new Karlsruhe Forschungsfabrik® (Karlsruhe Research Factory) is a manufacturing, development and test environment for this purpose. Together with AI experts, employees from companies and scientists from the institutes involved can creatively think up new ideas, develop them, and implement and improve them as prototypes up to implementation on an industrial scale in the trial areas of factory halls, free from the constraints of day-to-day operation. AI-based process optimisation Manufacturing companies now use artificial intelligence (AI) and machine learning (ML) to predict unexpected behaviour of machines or components during production and therefore avoid machine downtimes. Some companies are already developing models to predict quality, which are utilised to improve process runtime parameters. Over and above the optimisation of existing processes, AI can however help to rapidly bring new production methods to the stage of industrial maturit y. Considering the present challenges, e.g. market fluctuations, or an ever-increasing number of product variants, it is difficult for engineers to fully specify production processes in advance. They used to develop these processes, and conceived factories which were planned, configured, assembled, and commissioned. Nowadays, these steps often run parallel to each other. We use the term “immature processes” for production processes which have not yet been fully optimised, either because they utilise new methods, process new materials, or complex interactions occur between the raw material, the state of the process, and external influences. The idea behind AIbased rapid maturation is to implement processes and manufacture products on an industrial scale at a very early stage. Initially, high scrap rates are accepted if this rapidly produces adequate data for subsequent process optimisation. By means of (over)-instrumentation with extended use of sensors and actuators, the process is initially made accessible to machine learning methods. Karlsruhe Research Factory Bild: © Daniel Vieser, Architekturfotografie, Hildesheim/Karlsruhe Informatik in der Fabrik Die Welten wachsen zusammen. Ein Überblick by Olaf Sauer and Thomas Usländer EPUB 978-3-932298-96-7 PDF 978-3-932298-95-0 More info about the eBook https://t1p.de/adocv Video: Author live talk about the eBook German https://t1p.de/8n0qx