A Luenberger-type observer for the AM2 model

• We designed a nonlinear observer for the AM2 model with linear errors.
• We proved that this observer is robust with a global convergence and adjustable speed.
• For large speed of convergence, we noticed that in a small time interval at the beginning, some variables can take erroneous values (<0).
• The quantity of methane produced by the system can be estimated from variables x1 and x2 which qualifies our observer as a software sensor for the methane.

In this paper, we deal with the problem of designing a new observer for bioreactor models. The main idea is to construct a nonlinear observer with linear errors, which has an adjustable and robust convergence. Simulation results are presented using a model of Chemostat and a model of an anaerobic digestion process for the treatment of wastewater.