direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Page Content

Modelling, optimization and control of biological processes

Biological processes show complex internal regulation mechanisms and strong nonlinear behaviour which makes it difficult to find an appropriate control concept. Classical approaches based on heuristics or linearized models often yield insufficient results for biological systems. Therefore, the need for nonlinear model-based algorithms for process development, control and monitoring has increased.

Schauglas Fermenter

The quality of model-based concepts directly depends on the accuracy of the model and its parameters. A human modeller today mainly builds up the model manually in an iterative and therefore time-consuming way. Thus a software-tool TAM-B (Tool for the Automatic Modelling of Biological reaction systems) was developed, which automates the main steps of the modelling process. In case that there are only few experimental data available qualitative modelling is advantageous. Here methods like rapid prototyping are developed. Even if quantitative modelling is possible, often more than one model candidate can describe the measurements and/or model parameters are known imprecisely. Here methods of optimal experimental design for model discrimination and parameter identification have to be used for planning of experiments which supplement the experimental information's optimally. Using mathematical models, accurate monitoring by model-based state estimation algorithms and closed-loop control are possible. Because biological processes show strong nonlinear behaviour, classical linear control concepts yield insufficient results. Thus for closed-loop control adaptive controller and model-based control concepts like the Nonlinear Model Predictive Control or the Online Trajectory Planning are enhanced.

  • Automatic modelling of reaction systems, TAM-B (J. Leifheit)
  • Rapid prototyping for biological process development and control (N. Violet)
  • Optimal experimental design for model discrimination and parameter identification (M. Kawohl, T. Heine, N. Violet)
  • State estimation using Extended-Kalman-Filter, Constrained-Extended-Kalman-Filter and Moving-Horizon-State-Estimation (T. Heine, M. Kawohl)
  • Model-based control using structured compartment models including up to 16 states and 65 parameters (T. Heine, M. Kawohl)
  • Internet/wap-based supervisory monitoring and remote control of fermentation plants (M. Kawohl, T. Heine)

Experimental validation of the developed methods in real fermentations of e.g. Saccharomyces cerivisiae, Streptomyces griseus, Streptomyces tendae, Paenibacillus polymyxa and Ashbya gossypii is performed in our biolab. We have two 15L laboratory-scale fermenters integrated in a automatic process control system with data acquisition and storage, intenet/wap-based visualisation and supervisory remote control. Analysis of the nutrients, dry biomass, DNA, RNA and proteins is possible. (M. Valentin, Ch. Lange, H. Donner-Broszio)
Also there are mechanical experiments to validate statistical characteristics of the developed methods.

To top

Zusatzinformationen / Extras

Quick Access:

Schnellnavigation zur Seite über Nummerneingabe