Integrating Deep Residual Learning with Monod Kinetics for Bioprocess Control
Poster
| Status: | |
| Autorzy: | Król Krzysztof, Kulisz Monika, Kłosowski Grzegorz, Rymarczyk Tomasz, Wójcik Dariusz |
| Rok wydania: | 2026 |
| URL do źródła | LINK |
| Język: | angielski |
| Źródło: | 2026 ACM/IEEE International Conference on Embedded Artificial Intelligence and Sensing Systems - Posters and Demos (SenSys-Adjunct) |
| Miasto wystąpienia: | Saint-Malo |
| Państwo wystąpienia: | FRANCJA |
| Efekt badań statutowych | NIE |
| Abstrakty: | angielski |
| This study demonstrates a built-in bioreactor control system combining EIT with the ResNet-50 network. This model, which outperforms classical reconstruction methods, has been integrated with Monod kinetics within the “Digital Twin” framework to predict substrate consumption and biomass growth in real time. Validation during 46 hours of fermentation (SSIM 0.48, Pearson 0.74) confirms the reliability of the system in automatic decision-making. |
