Mathematical modelling of chemical processes—obtaining the best model predictions and parameter estimates using identifiability and estimability procedures

TLDR

Chemical engineers developing fundamental models often face difficulties estimating all parameters because of issues with parameter identifiability and estimability. The paper reviews the concepts of identifiability and estimability and discusses techniques, including MSE‑based methods, for selecting a subset of parameters to estimate. Reducing the number of estimated parameters can improve model predictions, yielding lower mean‑squared error. © 2011 Canadian Society for Chemical Engineering.

Abstract

Abstract Chemical engineers who develop fundamental models often have difficulties estimating all model parameters due to problems with parameter identifiability and estimability. These two concepts are reviewed, as are techniques for assessing identifiability and estimability. When some parameters are not estimable from the data, modellers must decide whether to conduct new experiments, change the model structure, or to estimate only a subset of the parameters and leave the others at fixed values. Estimating a reduced number of parameters can lead to better model predictions with lower mean squared error (MSE). MSE‐based techniques for parameter subset selection are discussed and compared. © 2011 Canadian Society for Chemical Engineering