Abstract:Hide melting presents itself as one of the most critical processes in the production of donkey-hide gelatin. Here a NIR-based method was established for the rapid analysis of in-process hide melting solutions as well as for end-point determination of this process. Near infrared (NIR) spectra of hide melting solutions were collected in transflective mode. With the contents of total nitrogen determined by the Kjeldahl method as reference values, partial least squares regression (PLSR) was employed to build calibration models between NIR spectra and total nitrogen. Model parameters including wavelength range and PLS factors were optimized to achieve best model performance. Based on the contents of total nitrogen predicted by calibration model, end point of hide melting was determined. The constructed PLS model gave a high correlation coefficient (R2) of 0.991 3 and a root mean square error of prediction (RMSEP) of 0.807 g·L-1. With the predicted total nitrogen and predefined limit, decisions concerning the proper times of melting were made. This research demonstrated that NIR transflectance spectroscopy could be used to expeditiously determine the contents of total nitrogen which was subsequently chosen as the indictor for determining the end-point of hide melting. The proposed procedure may help avoid unnecessary raw material or energy consumption.