EKI – Engineering for AI-based Automation in Virtual and Real Production Environments

Both end customers and industry need new products and product variants at ever shorter intervals. Manufacturers are not only faced with such time-to-market requirements, but are also under constant pressure to increase productivity, save resources and reduce costs. All of this has led to a strong need for new adaptive and changeable systems and in particular for corresponding automation solutions. Engineering in particular, i.e. the adaptation of the automation solution by experts, has proven to be a bottleneck, which is why engineering environments for automation solutions have to be reinvented.

Artificial intelligence (AI) is key to addressing these challenges. Currently, AI and especially machine learning (ML) are creating completely new approaches to automation, especially for adaptive, i.e. changeable systems (e.g. for new product variants). These AI approaches automatically allow the production sequence, i.e. to create the automation strategy or to adapt it to new requirements — instead of the previous laborious manual work. Other AI components such as predictive maintenance or energy analyzes use machine learning to monitor systems.

While such approaches to adaptive production have been called for on the European agenda for years, only a few individual aspects such as system modularization, software and mechatronic components or parameter optimization have been examined so far. No general solution has yet been developed for the core problem. This is mainly due to the problem of the lack of engineering methods:

From the user’s point of view, the engineering approaches for such adaptive systems are a key challenge: On the one hand, it is about being able to react faster to new requirements than before. On the other hand, ML and AI algorithms will also be permanently available as software components. However, such software components have to be put together in an engineering environment by integrators and machine builders (i.e. by people not from the AI ​​area) to form overall solutions. I.e. The core problem is not the algorithmic creation of such KI / ML components but the engineering environment. Corresponding engineering approaches do not exist today. As a solution, a new, open engineering approach with open interfaces is being developed in this project, which enables the integration of ML and KI-SW components into the engineering. A key idea are assistance functions to support plant operators in quickly creating new automation solutions.

Duration: September 1, 2020 to August 31, 2024

The EKI project is funded by the Federal Ministry of Defense (BMVg) in the DTEC funding program.


Letzte Änderung: 9. October 2020