The concept of modularizing plants is becoming increasingly important in the process industry, as it enables flexible manufacturing of various products. However, retrofitting or recommissioning modules leads to downtime periods in which no production can take place. These interruptions are associated with high costs, as no value creation is generated during downtime.

During these downtime phases, there is often no possibility to specifically optimize the respective plant components for the specific application case. This affects almost all commissioning, initial commissioning, and recommissioning processes. 
As a result, considerable optimization potential regarding the operating mode and control behavior of the plant is left untapped, which shortens the service life of the overall plant or individual components.
To address this challenge, the control systems should be enabled to reflect on their own behavior and tasks and continuously optimize them. Concepts such as Physical AI or Embodied AI make it possible to detect changes in the plant, autonomously derive decisions, and execute corresponding actions. In this way, both process performance and the longevity of the plants can be sustainably improved.

GOALS:

• Investigation of Physical and Embodied AI in the process industry
• Evaluation of AI algorithms taking into account process engineering conditions
• Implementation of selected AI concepts in a laboratory setup
• Evaluation of the approaches in the laboratory setup

HELPFUL PRIOR KNOWLEDGE:

• Interest in automation technology
• Enthusiasm for AI/ML (Python, TensorFlow/PyTorch)
• Lectures IT/AT, MARO and CPM