Development of a Dynamic Gray?Box Model of a Fermentation Process for Spore Production

Fermentation processes are difficult to describe using purely mechanistic relations as the underlying biochemical phenomena complex and often not fully understood. In order cope with this challenge, we developed an approach augment standard dynamic model equations by data-based components that fitted data machine learning techniques, which results in gray-box models. This methodology is applied here batch fermentation process of sporulating bacterium Bacillus subtilis, experimental from a lab-scale fermenter. The key step developing estimation training set for submodels. quality resulting analyzed, predictions compared real data.