Calibration Modeling for On-line Monitoring of Ethanol Fermentation Processes Based on FT-NIR Spectra Technology

Aiming at the on-line detection of glucose concentration, ethanol concentration, and biomass in ethanol fermentation process, we propose a spectral calibration model building method for measuring these parameters on-line by the Fourier transform-near infrared (FT-NIR) spectroscopy technology. We adopt the partial robust M-regression (PRM) method to eliminate the influence of collected spectral outliers to model calibration and present a grid-searching method to determine the optimal component number as well as the weighting coefficients for modeling. Moreover, we evaluate the model validity in terms of accuracy, stability, and resolution. The results show that the model established by the PRM method has better prediction effect. Finally, an on-line monitoring experiment for the ethanol fermentation process is performed to verify the effectiveness of the proposed calibration model and on-line monitoring method.