Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Forschung --- Zeitschrift --- Forschungsbericht --- Otto-von-Guericke-Universität Magdeburg

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Determination of bioreactor operating policies for optimal performance of a bioprocess is a challenging task due to the highly variable nature of biological systems and our limited process knowledge. To address this challenge, model-based optimization and control approaches can be implemented for conducting in silico experiments to derive optimal control strategies for improved performance of bioprocesses.For reliable performance of model-based optimization and control, it is crucial that the underlying model provides proper levels of detail to represent the real bioprocess and to address the full metabolic versatility. In this work, we consider bioprocess optimization and control by exploiting the capabilities of dynamic metabolic-genetic network models. In particular, we consider improving bioprocess productivity through temporal manipulations of metabolism using dynamic enzyme-cost FBA model (deFBA). The dynamic nature of this model and included details on gene level allow for direct temporal manipulation of gene expression, and through a proper formulation (a bilevel problem), one can identify optimal genetic and process level manipulation strategies according to the target performance criterion (productivity). Moreover, advanced bioprocess control and optimization requires flexible and robust control strategy which guarantees the performance of the model-based approach in the presence of disturbances and existing uncertainties. To this aim, on-line adaptation schemes are integrated within our modeling approach which are suitable to control highly uncertain biological processes with fast reactions to disturbances. The adaptive approach could allow for online adaptation of the underlying model (deFBA) by estimating uncertain and variable model parameters in different stages of the process. In this direction, the developed deFBA-based approach is implemented inside a model predictive control (MPC) routine, combined with a moving horizon estimation (MHE) algorithm in order to adjust the underlying model online for different metabolic modes.Considering the case study of ethanol formation in E. coli under different growthconditions, it is shown that the proposed approach is a suitable approach to optimizeand control time-varying bioprocesses. Desired engineering objectives can be addressed by the proposed approach through temporal manipulations of the metabolism while process uncertainties can be handled efficiently using the adaptive nature of the implemented control scheme.