There are often substantial differences between model-based planning data of process plants and the process data actually recorded during plant operation, and clarifying these differences is highly relevant for the further development of the processes.
The seemingly obvious use of standard machine learning methods is not effective here, since the complex sensors and analytics in production plants usually only incompletely capture the input-output behavior and can only be learned to a limited extent based on this.
Therefore, the essential goal is to close such systematic gaps by means of models through the strongly interdisciplinary cooperation in the transfer center »Process Engineering / Chemistry« and thus to advance decision support and process development in the chemical industry.