FideLio – Predictive fouling detection in food processing by models based on real data
Funding: German Federal Environmental Foundation (DBU)
Funding code: 37305/01
Project term: 01.02.2022 – 31.07.2024
Contact: Niklas Jarmatz, M. Sc.
Industrial food production is characterized by the high use of resources (energy, water, cleaning agents) for the processing of the food stuff and the subsequent cleaning of the production facilities. Due to the high degree of automation and the high level of security provided for the hygienic implementation of the cleaning processes, the savings potential is correspondingly large.
The overarching goal of the project is to set up a self-sufficient overall system for predicting the fouling behavior in food production. For this purpose, a combination of commercially available clamp-on sensors for process monitoring, data acquisition via an IoT-system and software for data preparation and evaluation with an AI component will be created. In the end, the demonstrator should be able to be integrated into new systems with little effort as a complete device with hardware and software.
In addition, an aim is to provide a holistic methodology for predicting the fouling behavior of critical plant components in food production. Previous experience in this area and large amounts of data collected in the industrial practice will be used for this purpose. From this data, models are developed using machine learning methods, which can detect fouling reliably and at an early stage.