Rapid measurement of antioxidant activity in dark soy sauce by NIR spectroscopy combined with spectral intervals selection and nonlinear regression tools

This paper attempted to show the feasibility of measuring the antioxidant activity in dark soy sauce by NIR spectroscopy technique. Chemometrics on spectral intervals selection and nonlinear regression tools were systematically studied in the calibrating model. First, the optimal spectral intervals were selected by synergy interval-partial least square (Si-PLS). Then, kernel PLS (KPLS) and back propagation artificial neural network (BPANN), as two nonlinear regression tools, were performed comparatively to calibrate models based on optimal spectral intervals, called Si-KPLS and Si-BPANN models, respectively. These models were optimized by cross-validation, and the performance of the final model was evaluated according to correlation coefficient (R-p(2)) and root mean square error of prediction (RMSEP) in the prediction set. The results showed that the Si-BPANN model was superior to other models, and the optimal result was achieved with R-p(2) = 0.9769 and RMSEP = 0.0221 in the prediction set. This work demonstrated that total antioxidant capacity in dark soy sauce could be measured by NIR spectroscopy technique, and Si-BPANN showed its superiority in model calibration.