Endogenous model state and parameter estimation from an extensive batch experiment

In this paper an extensive batch experiment of endogenous process behavior in an aerobic biodegradation process is presented. From these experimental data, comprising measurements of MLVSS (mixed liquor volatile suspended sol ids) and respiration rate, in a first step the states and unknown parameter s in a four-compartmental model are reconstructed analytically. Subsequentl y, for a selected set of states and parameters, using the results of the pr evious step, a recursive state estimation procedure, in particular an Exten ded Kalman filter-based observer, is applied to deal with the noise propert ies of the data appropriately. From this it appears that the initially prop osed model structure, and especially the hydrolysis term, has to be modifie d. (C) 2000 John Wiley & Sons, inc. Biotechnol Bioeng 68: 422-429, 2000.