Heuristic parameter estimation for a continuous fermentation bioprocess
Zymomonas mobilis continuous fermentation bioprocess has the ability of producing energy from glucose catabolism, which promises a relevant application for biomass conversion into fuel, and therefore it represents an industrial scale production alternative for our country. However, it has demonstra...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Universidad de Antioquia
2018-09-01
|
| Series: | Revista Facultad de Ingeniería Universidad de Antioquia |
| Subjects: | |
| Online Access: | https://revistas.udea.edu.co/index.php/ingenieria/article/view/329055 |
| Summary: | Zymomonas mobilis continuous fermentation bioprocess has the ability of producing energy from glucose catabolism, which promises a relevant application for biomass conversion into fuel, and therefore it represents an industrial scale production alternative for our country. However, it has demonstrated high complexity regarding the non-linear and non-Gaussian characteristics of its dynamics. Several works have been dealing not only with the bioprocess modeling but also with controller design and implementation. These works have developed state and parameter estimation strategies based on particle filters and Gaussian methods, as well as closing the loop with nonlinear controllers. Nevertheless, there is a need to improve previous parameter estimation results, enabling future design of control strategies for industrial applications. We present a set of heuristics algorithms for the non-linear system parameter estimation evaluated with data from 150 hours of fermentation. Some algorithms such as local search methods, simulated annealing, population heuristics, differential evolution, bacterial chemotaxis and other techniques were tested for the bioprocess. Simulations of the microorganism model and experimental verifications showed the good performance in parameter accuracy and convergence speed of some of the heuristic methods proposed here. Moreover, the reliability and acceptable computational costs of these methods demonstrate that they could also be applied as parameter estimators for other bioprocesses of a similar complexity.
|
|---|---|
| ISSN: | 0120-6230 2422-2844 |