Application Studies On Rapid Analysis Of HTPB Based Solid Propellant Slurry By NIR Spectroscopy

The determination methods of dioctyl sebacate (DOS), burning agent (BA) and ammonium perchlorate (AP) in hydroxyl terminated polybutadiene (HTPB) based solid propellant slurry were investigated by means of near infrared (NIR) technique. The effects of chemometrics methods which were applied in spectral data pretreatment, calibration model establishing and outlier detecting ware discussed. A new approach for quantitative analysis of main components in HTPB based solid propellant slurry was provided.In part Ⅰ, the feasibility quantitative analysis on HTPB propellant slurry by near infrared spectroscopy was discussed, while both instrument error and artificial error are within the acceptable range. Regression models on the concentration of DOS in samples of two components system were investigated by methods of peak area internal standard linear regression and partial least squares (PLS). Both of them obtained the correlation coefficient reached more than0.95, proved that the near infrared spectroscopy is feasible in quantitative analysis of HTPB propellant slurry.In part II, the quantitative analytic models of DOS and BA in three components slurry samples were established using the method of PLS. The model was optimized by means of Mahalanobis distance and principal component analysis (PCA). Compared the predictive ability of two optimized models and the simplicity of operation, the method of principal component analysis is more suitable for the model. The absolute error of predictive value and actual value of DOS and BA was2.010%and1.619%, and the relative mistake of predictive value and actual value of DOS and BA was0.0815and0.1029by the optimal model.In part III, the quantitative analytic models of DOS and BA in six components slurry samples were established using the methods of PLS-1and PLS-2. Compared the prediction results of PLS-1and PLS-2, both of the two models could predict the contents of DOS and BA in the known sample accurately. PLS-2model was more suitable for the propellant slurry samples due to the characteristics of fast and convenient in the modeling and forecasting process. The absolute error of predictive value and actual value of DOS and BA was0.399%and0.304%, and the relative mistake of predictive value and actual value of DOS and BA was0.0371and0.0424by the PLS-2model.In part IV, the quantitative analytic models of AP in seven components slurry samples were established using the method of PLS. The absolute error of predictive value and actual value of AP was0.569%, and the relative mistake of predictive value and actual value of AP was0.0084by the model. But the concentration of AP was just from60%to80%in this experiment and the model would have some limitation in the practical application. The modeling data should be added to expand the scope of the model in the future.