Rapid determination of the moisture content of triethyleneglycol dinitrate absorption tablets by near-infrared spectroscopy

Abstract

During the production of triethyleneglycol dinitrate (TEGDN) double-based propellants, excessive moisture content adversely affects the plastination of TEGDN absorption tablets. This study focused on developing a model for detecting moisture content in the TEGDN absorption tablets by near-infrared spectroscopy (NIRS). The spectral intervals of 1149.7–1248.8 nm and 1397.4–1515.1 nm were determined according to the competitive adaptive reweighted sampling (CARS) algorithm and comparison of the absorption peaks of TEGDN absorption tablet samples and the absorption peak of water. A combination of standard normal variables transformation (SNV) and first-order derivative (FD) was selected as the original spectral pre-processing method. The optimum number of factors for the moisture model was chosen as 7. A moisture quantification model was developed based on the Partial Least Squares (PLS) algorithm. The determination coefficient of the calibration and cross-validation (Rc2,Rcv2) were 0.9905 and 0.9869, respectively. The root means square error of the calibration and cross-validation (RMSEC, RMSECV) were 0.0158 and 0.0186, respectively. The developed moisture model was externally validated using a prediction set. The determination coefficient of the prediction (Rp2) was 0.9899, and the root mean square error of the prediction (RMSEP) was 0.0207. The mean absolute and mean relative errors between the predicted values of the NIRS method and the measured values of the traditional method were 0.0091 and 1.0819%, respectively. Therefore, the results show that the moisture model developed by NIRS can quickly and accurately detect the moisture content of TEGDN absorption tablets after removing moisture and determine whether the TEGDN absorption tablets are qualified.