Trade-offs in design of fed-batch experiments for optimal estimation of biokinetic parameters

摘要

The estimation of biokinetic parameters from experimental data is known to be a tedious step in the identification of bioprocess models. Optimization of a given feed rate profile structure has been performed with respect to a novel combined criterion aiming at both decorrelation and high accuracy of the parameter estimates. For specific values of a weighting factor, the cost reduces to the modified E-criterion, the E-criterion, or the D-criterion for optimal experimental design. It is illustrated how this novel criterion leads to trade-off problems in parameter estimation, such as quality of the parameter estimates, practical feasibility, and model validity.