Optimal experimental design for the calibration of models of phosphorus removing activated sludge systems

摘要

In the field of waste water treatment, the Activated Sludge Models have been introduced as a tool for upgrading, optimising and controlling the process. The identification of these models is in general characterised by two important features, i.e. the models are often highly complex and there is a lack of reliable sensors for on-line measurement The structural and practical identifiability of the parameters in these models is addressed in this paper. It focuses on optimal experimental design of experiments (OED), i.e. procedures to design a limited number of experiments that give the experimenter the maximum information content. The experimenter will select the experimental region and perform one experiment. From then onwards following experimental conditions will be chosen from the experimental region solely based on their predicted information content. The method is applied to calibrate models of phosphorus removing activated sludge systems. The D-criterion was found to give the best performance for the selection of the experimental conditions.