Chemometric modelling for process analyzers using just a single calibration sample

A new calibration method for process analyzers, which requires a chemometric model, is presented. Instead of a huge amount of calibration data, which are usually necessary to calculate a chemometric model, just a single off-line sample is necessary for the proposed procedure. The further required information to calculate the chemometric model is made available in a form of a theoretical process model, whose parameter values will be estimated inherently during the calibration procedure. The functioning of the new procedure is demonstrated by its application to 2D-fluorescence data for the monitoring of a biotransformation process. In the process acetophenone is converted to (S)-1-phenylethanol.Using the new calibration method the relative errors of prediction of the substrate and product concentrations are 4.9% and 4.6%, respectively.