Premium
Adaptive nonlinear cell mass state estimator for a continuous yeast fermentation
Author(s) -
Ramseir Michael E.,
Agrawal Pramod,
Mellichamp Duncan A.
Publication year - 1993
Publication title -
aiche journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.958
H-Index - 167
eISSN - 1547-5905
pISSN - 0001-1541
DOI - 10.1002/aic.690390214
Subject(s) - nonlinear system , estimator , robustness (evolution) , control theory (sociology) , a priori and a posteriori , state variable , adaptive estimator , mathematical optimization , computer science , biological system , mathematics , control (management) , statistics , chemistry , biochemistry , physics , philosophy , epistemology , quantum mechanics , artificial intelligence , biology , gene , thermodynamics
On‐line measurement of the important state variables in fermentations, particularly cell mass concentration, remains a difficult problem. However, a number of secondary or environmental variables can be measured conventionally and on‐line, such as pH, and CO 2 and O 2 in the exhaust gas. Stephanopoulos and San (1984) have developed a modeling approach, based on species balances, that provides relations between the environmental and important state variables. Using such a model, the important state variables can be estimated in principle from more easily accessible on‐line measurements. In this article, a new adaptive estimator is developed, incorporating as its basis an underlying nonlinear model so as to utilize the best possible a priori process knowledge. Base addition rate and CO 2 offgas concentration are measured on‐line and periodically. Cell mass measurements are incorporated infrequently and even at irregular sampling periods, thus providing a very flexible scheme. Only a single adapted parameter is required to match the model to the plant operating characteristics. This simple but rigorous model form results in an estimator that is easy to implement and to tune and which exhibits long‐term robustness due to its multirate feedback structure. Experimental results from a laboratory‐scale continuous fermentor show that such a cell mass estimation scheme yields excellent performance both open‐loop (without control) and as a part of conventional and nonlinear adaptive control approaches.