KI-BAYOPT - AI-assisted online optimization for highly efficient process control in Bavarian plant engineering

The global trend towards digitalization and automation of manufacturing processes and operations monitoring in the industry leads to more efficient design and optimization of processes through the use of analysis and monitoring tools. While mass production generates large amounts of data that facilitate the training of AI models, the manufacturing of customized machines and equipment is characterized by small product quantities, long manufacturing time, high degree of manual labor, and correspondingly low process automation capability. The small batch sizes mean that only limited data is available, so that established machine learning approaches are not transferable.

As part of the KI-BAYOPT project funded by the Bavarian Research Foundation (BFS), an AI-based predictive model is being developed for analysis and decision support for operations monitoring, taking into account production and manufacturing processes for small batch sizes. Using the production and performance of helium transport containers as an example, the predictive analytics model will improve process management by combining real-time measurement data with expert knowledge, historical data, and simulation models to make statements about process stability and derive recommended actions for workers.