Main Page

Project Wiki and Semantic Dataspace based on OpenSemanticLab for the KIproBatt Cell Manufacturing (public) and Project Management (private).

To access the JupyterLab Workflow Environment, visit kiprobatt.de/jupyter (private)

Git Repos exist both on Github (archived dataset) and FhG Gitlab (data & AI tools).

For citing the platform, please use https://doi.org/10.1002/ente.202301305.

If you reuse our published data, consider citing our data repository: https://doi.org/10.5281/zenodo.8418323 .

What can I do here?

• 📖

  Glossary

  Take a look at our interactive glossary - especially the Cell manufacturing

  Open Glossary
• 🔁

  Process Spec

  Explore our Semantic Process Definition - from the top down to single parameters

  Show Spec
• ⟺

  Correlations

  Follow the correlations we have classified as important

  View Map
• ❉

  Production Knowledge Graph

  View the Knowledge Graph of a single produced cell

  Show Graph
• ❉

  Analytics Knowledge Graph

  View the Semantic Metadata Structure of an AI based data analysis

  Show Graph
• 📊

  KPIs

  Inspect our Key Performance Indicators (KPIs)

  Show KPIs

About KIproBatt - Intelligent battery cell manufacturing with AI-supported process monitoring based on a generic system architecture

Funding authority:

BMBF InZePro Cluster

Project duration:

10/1/2020 – 3/31/2024

Partners:

• Fraunhofer Institute for Silicate Research ISC
• Fraunhofer Institute for Casting, Composite and Processing Technology IGCV
• University of Bayreuth
• Aschaffenburg University of Applied Sciences (UAS)

Contact:

• Simon Stier: simon.stier@isc.fraunhofer.de

Abstract:

Due to its complexity and vast economic as well as ecological impact, the Li-ion battery cell production process is subject to ongoing digitization and optimization in order to increase cell performance while reducing resource consumption and production costs.

In this context, artificial intelligence (AI) holds immense potential in leveraging manufacturing data to improve the cell production process. Hence, we aim to enhance cell production with AI-based end-to-end process monitoring, which covers all steps of the process chain. For this purpose, we develop a generic, software-implemented system architecture as reusable structure that allows us to connect the process data acquisition with a carefully constructed ontology-based semantic data space. Based on this system architecture, we attach machine learning approaches from two perspectives: In the first perspective, we apply both data and physics driven models to specific process parameters to detect and evaluate correlations and process anomalies. In the second perspective, we develop an overarching end-to-end process monitoring. For this application, we integrate the previously developed models into a dashboarding system in order to assess their relevance for cell performance and monitor the cells' state in production constantly and as real-time as possible.

The combination of these two perspectives allows us to detect defects early in the production process, rapidly increase the quality performance and derive flexible adjustments to the process parameters in case of malfunction or defects. Thus, we expect to reduce the total cost of cell production as well as improve its carbon footprint by reductions in resource and energy consumption. Finally, the generic, semantically structured design enables an easy transfer to other research and industrial processes.

About the KIproBatt Dataspace

Technically the KIproBatt plattform consists of the following OpenSource components

• MediaWiki: Most important Wiki framework, basis of Wikipedia
• Semantic MediaWiki: Semantic extension for MediaWiki
• JupyterLab: Cloud-based scripting environment, e. g. for Python

MediaWiki provides the basic document-based data structure. Furthermore, extensive interfaces for human (editors) and machine (API) are provided. Templates and forms ensure uniformity and efficiency. The extension Semantic MediaWiki provides the Wiki with numerous possibilities for linking and annotating data. In addition, semantic queries can be sent directly to the Wiki and results can be visualized in a variety of ways. Via JupyterLab users can access data stored in the Semantic MediaWiki, process it and push the results back to the Wiki.

This project is funded by the German Federal Ministry of Education and Research (BMBF), grant no. 03XP0309A-C.