Some of our latest research has been published in Systems & Control Transactions under the title “A Physics-Informed Approach to Dynamic Modeling and Parameter Estimation in Biotechnology”. This study explores the use of physics-informed neural networks for bioreactor dynamic modeling.
Konstantinos Mexis, Stefanos Xenios, Nikolaos Trokanas and Antonis Kokossis explore how these advanced hybrid methods can overcome significant challenges of bioprocess dynamic modeling. By integrating fundamental mechanistic laws with deep learning, this approach provides a robust framework for developing digital twins for bioreactors.
This integration of machine learning and scientific principles represents a significant step toward more intelligent, data-driven solutions for industrial biotechnology, aligning with the objectives of the Bioindustry4.0 project.
You can read the full article here.