INFERRING CONCENTRATIONS ONLINE FROM NEAR-INFRARED SPECTRA - NONLINEAR CALIBRATION VIA MIDINFRARED MEASUREMENTS

Near-Infrared (NIR) spectroscopic methods for on-line concentration es timation are gaining popularity in chemical production. The problem wi th NIR data, however, is its nonlinear nature rendering standard facto r-analytical (FA) techniques inapplicable to estimate the concentratio ns. To infer concentrations on-line from NIR spectra, the following st eps are proposed: (i) measure both the NIR and Mid-Infrared (MIR) spec tra of unknown reaction mixtures in the laboratory (MIR spectroscopy h as the desirable property of linearity with respect to concentrations) , (ii) apply FA techniques to MIR data and infer concentrations, (iii) calibrate NIR against the estimated concentrations by using standard nonlinear regression methods and (iv) use the calibration model in sub sequent production runs. Experimental results are presented to illustr ate the efficiency of this method.