PROMICON’s fourth Graphical Summary: New insights into bioreactor systems

PROMICON’s initiative to share graphical summaries for our research papers aims to make complex scientific studies more accessible and engaging for interested stakeholders such as policymakers, the scientific community, and the general public. These summaries will be available on our website Media Center and social media platforms, providing readers with a quick and informative overview of our research highlights.

We continue with the paper “A General Deep Hybrid Model for Bioreactor Systems: Combining First Principles with Deep Neural Networks”. This study is aimed at enhancing bioreactor modeling by integrating deep neural networks with First Principles for improved predictions and generalisation. Some of the key findings of the study include:

• Deep hybrid models: A novel paradigm combining the complexity of deep neural networks with the rigor of First Principles equations;

• Advanced and efficient training: 43.4% faster training compared to traditional shallow models;

• Real-world applications: Tested on synthetic data and a pilot 50L bioreactor;

• Accuracy boost: 18.4% increase in prediction accuracy over shallow models.

These advancements offer significant value for bioprocess engineers and researchers, providing a deeper understanding and control of bioreactor systems.

Read more here and stay tuned for more graphical summaries.

Snippet from PROMICON’s fourth Graphical Summary