| Résumé | A multi-model observer based estimator (OBE), which consists of a set of simple sub-models describing typical process states and a knowledge-based system, was developed for monitoring and long-term prediction of the anaerobic digestion process. The knowledge-based system uses biogas and pH measurements for process diagnosis, while OBE uses fluorescence-based measurements of chemical oxygen demand (COD) and volatile fatty acids (VFAs) for estimating process outputs. The approach was tested on experimental results obtained from the operation of a lab-scale anaerobic reactor. The sub-models used either Monod or a combination of zero- and first-order rate expressions. A comparison showed that while both multi-model OBEs featured rapid convergence and good stability, the combined zero- and first-order kinetic equations provided better long-term prediction accuracy. |
|---|
