Quickly determination of resorcinol in composite modified double-base propellants based on near-infrared reflectance spectroscopy

• We first developed the NIR prediction model for determination the resorcinol content. • The optimum model is achieved with modeling interval of 4126.92–6927.05 cm −1 . • SNV + 1 ST is the appropriate spectral pretreatment method for resorcinol prediction model. • The prediction model of resorcinol enables to judge whether the sample's stability is qualified or not. Resorcinol, as a stabilizer component, plays an important role in maintaining the stability of composite modified double-base (CMDB) propellants. Near-infrared reflectance spectroscopy (NIRS) was used to quickly detect the resorcinol content of CMDB propellant. The optimal spectral interval was determined to be 4126.92–6927.05 cm −1 by analyzing the useful characteristic peaks in sample spectra. The partial least squares (PLS) algorithm was used to establish the resorcinol prediction models. By comparing the performance of the developed models with different preprocessing, the best method of spectral preprocessing was determined as standard normal variate transformation (SNV) + first-order derivative (1 ST ). The optimal number of factors was 7 for resorcinol model. For the developed model, the determination coefficients of calibration and cross-validation ( R c 2 , R cv 2 ) were 0.997 and 0.987, respectively. The root mean square errors of calibration and cross-validation (RMSEC, RMSECV) were 0.013 %, 0.024 %, respectively. The determination coefficients ( R p 2 ) and the root mean square errors of prediction (RMSEP) of prediction were 0.991 and 0.018 %. The average relative error of the proposed and reference methods was only 0.253 %. The developed method can be used for quickly determining the content of resorcinol in CMDB propellants. In addition, methyl violet test showed that the stability of CMDB propellant became worse with the decrease of resorcinol content. In terms of stability, qualified CMDB propellant should contain resorcinol content not less than 1.285 %. Thus, the prediction model of resorcinol is available for judging the stability of CMDB propellants qualified or unqualified.