ManuBrain – Universal, scalable AI-platform for industrial applications
Fraunhofer SCAI exploits the potential of machine learning methods for industrial applications in SMEs
In the "ManuBrain" project, a universal, scalable, and open platform for artificial intelligence applications in medium-sized industrial companies is being developed. The state of North Rhine-Westphalia and the European Fund for Regional Development are supporting the project over three years with a total of 1.8 million euros. Fraunhofer SCAI develops and evaluates machine learning methods for engineering applications.
So far, only a few medium-sized companies have been using artificial intelligence (AI) methods to analyze sensor data in production plants. Therefore, the project »ManuBrain« aims to create an AI platform that helps to open up the potential of data analysis for concrete industrial applications. In ManuBrain, partners from research and industry are working together to ensure the practical relevance of the developments.
Fraunhofer SCAI contributes its long experience in data analysis and machine learning to the project. An important role is played by the so-called "informed machine learning". Here, application knowledge is also used in data analysis, for example in the form of physical assumptions. "An interdisciplinary cooperation of researchers and engineers is necessary to make application knowledge usable for machine learning (ML)," says Prof. Dr. Jochen Garcke, head of the business area "Numerical Data-Driven Prediction" at Fraunhofer SCAI.
For example, it is possible to investigate the influence of external conditions, such as temperature, on the properties of a product. Products and processes can be optimized by combining collected data, experience of the machine operator and physical-technical knowledge.
One focus of the research of Fraunhofer SCAI is on transfer learning methods. They can help to transfer machine learning models that work for one type of machine to related machine types. In addition, attention is paid to the interpretability of the results of the ML methods.
Project partners are the University of Siegen, Fraunhofer SCAI (Sankt Augustin), Recogizer Group GmbH (Bonn), Albrecht Bäumer GmbH (Freudenberg), SMS group GmbH (Düsseldorf) and ZENIT GmbH (Mülheim an der Ruhr).
ManuBrain is part of the lead market competition IKT.NRW, in which eleven innovation projects are being supported with around 16 million euros.