Loading...

Data-driven Modeling of a Fermenter Using Flux Balance Analysis

Banitalebi Dehkordi, Milad | 2022

190 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 55730 (06)
  4. University: Sharif University of Technology
  5. Department: Chemical and Petroleum Engineering
  6. Advisor(s): Pishavie, Mahmoud Reza; Vafa, Ehsan
  7. Abstract:
  8. Optimization and control of complex bioprocesses require accurate modeling. The application of flux balance analysis has gained considerable attention. In this approach, structured modeling is constructed based on metabolic genome networks. Although flux balance analysis provides an accurate model, applying it in optimization and control schemes especially model predictive control strategies often lead to bilevel optimization problems with significant computational cost. Therefore, we require a faster approach. The aim of this work is to use hybrid neural networks as a substitute for the optimization problem embedded in the dynamic model of a fermenter. In this work by proposing a hybrid model as a substitute for dynamic flux balance analysis, the practical optimization of maximizing ethanol production by saccharomyces cerevisiae as a case study was investigated. To this aim, using the developed hybrid model the optimal profile of ethanol was obtained, and a Laguerre model predictive control scheme coupled with a semi-Luenberger observer, as a state estimator, for optimal profile tracking has been incorporated. The results show that applying deep convolutional neural networks in a hybrid model structure is a suitable substitute for flux balance analysis-based models. It is also, shown that by using the prescribed hybrid model, the computational cost is reduced significantly and it is practically possible to implement online model predictive control strategies in bioprocesses
  9. Keywords:
  10. Flux Balance Analysis Method ; Hybrid Neural Network ; Hybrid Modeling ; Leunberger-Like Observer ; Laguerre Model-Based Predictive Control ; Bioprocessing ; Fermenters ; Data Driven Modeling

 Digital Object List

 Bookmark

...see more