Application Studies On Rapid Analysis Of Double-Base Oblate Spherical And HTPB Propellants By NIR Spectroscopy

The high-energy risk of propellants has a certain lag in the quality control level.The traditional methods have exposed the defects of inefficiency,cumbersome and high cost while employing them to ensure accuracy.Currently,the efficiency of traditional methods fails to keep pace with the development of production process,and thus they can’t meet the requirement of rapidly detecting the quality of product during the production process.Near-infrared（NIR）spectroscopy is a method in a fast,efficient,and environmentally friendly manner,which has become a novel approach to monitor the quality of product.The objective of this paper is to investigate the feasibility of rapidly monitoring the content of target component in the double-base oblate spherical propellant（DOSP）,and/or tracking the blend uniformity of hydroxy-terminated polybutadiene（HTPB）propellant in a continuous mixing process,with the help of using NIR spectroscopy.Based on the NIR spectral data from 40 DOSPs coupled with partial least squares（PLS）,a quantitative model was developed for rapidly detecting the content of ethyl acetate（DBP）in the DOSPs.The effects of pretreatment methods,variable selection and PLS factor number on the performance of the developed model were investigated.In this work,the determination coefficient of the calibration(²)and the root mean square error of cross-validation（RMSECV）of the developed model with the best performance were 0.9786 and 0.0536,respectively.Moreover,the determination coefficient of prediction(²)and the root mean square error of prediction（RMSEP）for DBP model were 0.9883 and 0.0412,respectively.The average deviation between the reference values and NIR predicted values for DBP was 0.017%;and the average relative error between the reference values and NIR predicted values for DBP was1.529%,at a relatively low level,confirmed that the established DBP quantitative model had a good predictive accuracy.Based on the NIR spectra from 42 DOSPs combined with PLS,a quantitative model was developed to monitor the content of DBP and N,N′‐dimethyl‐N,N′‐diphenylurea（C₂）in the DOSPs.The effects of pretreatment methods,variable selection and PLS factor number on the performance of the developed models were investigated.The²of DBP/C₂ models with the best performance were 0.9771 and 0.9840,respectively;RMSEC of DBP/C₂ models were0.0173 and 0.0072,respectively.²were as high as 0.9681 and 0.9554 for DBP/C₂ models,respectively;and RMSEP were as low as 0.0275 and 0.0107 for DBP/C₂ models,respectively,indicated that the deviation between the reference values and NIR predicted values was low.Additionally,the average deviations between the reference values and NIR predicted values for DBP and C₂ were 0.024%and 0.008%,respectively;and the average relative errors between the reference values and NIR predicted values for DBP and C₂ were 0.652%and 0.429%,respectively,at a relatively low level.It showed that the created DBP/C₂ quantitative models both had a high predictive accuracy.The NIR spectra of HTPB propellant slurries during the continuous mixing process were collected.The qualitative model was developed through orthogonal partial least squares discriminant analysis（OPLS-DA）to identify the blend uniformity of slurries during the mixing process.In this work,samples in the training set was divided into I,II groups according to the result of principal component analysis（PCA）model.The effects of modeling methods,preprocessing methods,variable selection methods on the performance of OPLS-DA models were investigated.The squared correlation coefficients（R²）of the OPLS-DA models with the best classificational performance for I-II groups were 0.806 and 0.980,respectively,and the predictive squared correlation coefficient（Q²）of the OPLS-DA model for I-II groups were0.737 and 0.975,respectively.In addition,the OPLS-DA models and Euclidean distance（ED）have relatively consistent results for judgement the blend uniformity of the slurries in each mixing stage,confirmed that the method we proposed was a potential tool to record the behavior trajectory of HTPB propellant slurries in the mixing process.The time for HTPB propellant slurries to reach uniformity was sucessfully estimated by OPLS-DA models.