Artificial intelligence optimizing processes in industrial companies

The use of machine learning makes it possible: knowledge is gained from data and introduced at all levels of the company's processes in order to generate added value. This allows design processes to be improved, product weaknesses to be identified and developments to be accelerated. The aim of the Machine Learning for Production and its Products (ML4Pro2) project is to make machine learning accessible to companies for intelligent products and production processes.

Together with a powerful consortium of industrial partners, Fraunhofer IEM is working in the project to optimize existing processes and work steps using machine learning methods. Several use cases are being used to evaluate the potential for industrial companies. Fraunhofer IEM is collaborating with Benteler and Hesse Mechatronics in the areas of predictive quality, process optimization and hybrid learning methods, and is developing intelligent assistance systems for production plants.

Such a system supports Hesse Mechatronics and its customers in the process-specific setup of their machines. The intelligent determination of process parameters by machine learning is intended to shorten this process of commissioning and help detect unknown sources of error.

The focus of the collaboration with Benteler is on predictive quality, the continuous monitoring of product quality through data-based predictions. By recording and evaluating production information, patterns in the data are identified and examined, on the basis of which product quality predictions can be derived. This will minimize waste, increase process reliability, and avoid product recalls.