Rapid Detection Method Of Solvent Content In Continuous Production Of Single Base Propellant

Single-base propellant is a kind of soluble plastic propellant with nitrocellulose as its main component.It is mainly used as the energy source of many kinds of ammunition and small artillery.The main technological processes of single-base propellant include water displacement,plasticizing and stretching,extrusion,herbal medicine cutting,pre-drying,soaking,drying and moisture absorption,mixing and packaging,among which solvent components play a vital role.Alcohol ether solvents in plasticizing and pressing process can gelatinize nitrocellulose,and solvent content can also reflect the mixing uniformity of single-base propellant,which is related to the compactness and compaction of finished single-base propellant.In the process of herbal medicine cutting,the content of alcohol and ether mixed solvents in single-base herbal strips plays a decisive role in the effect of herbal medicine cutting.The content of pre-drying and drying efflux in the process of dispelling solution and controlling efflux affects the shrinkage rate of single-base propellant and the combustion performance of finished propellant.At present,it takes more than 2 hours to detect the external swing content by weightlessness method for pre-drying process and drying process.However,due to the limitation of weightlessness method in plasticizing,pressing and herbal medicine cutting process,military factories can not be used to detect the solvent content,which results in only relying on the workers'sensory judgment.In this paper,near infrared spectroscopy?NIRS?is used to rapidly analyze the solvent content in each process.The specific research contents are as follows:Near infrared spectroscopy?NIRS?was used to determine the content of alcohol ether solvent in plasticizing and pressing process.By comparing the RMSECV values and comparing the root variance of the models established in different intervals,the best pretreatment method for the solvent content model in plasticizing and extrusion links was determined to be the first derivative+MSC,and the best modeling intervals were 7507.7-5447.4 cm^-1 and 4606.8-4244.2cm^-1.The optimum principal factor of the model of alcohol ether solvent content in the process of single-base propellant pressing was determined to be 7 by using the method of leave-one interaction validation.The correlation coefficient R_C² _V is 0.9855,RMSECV is 0.297,RPD is8.3,RMSEP is 0.289,RMSEP is 0.289,the range of prediction is within 1%,the standard deviation is below 0.4%,and the average prediction deviation of single sample is within 0.3%.Near infrared spectroscopy?NIRS?was used to determine the content of alcohol and ether solvents in the process of cutting herbs.By comparing and analyzing the near infrared spectra of single-base drug samples and nitrocellulose,alcohol and ether solvents,and comparing the root mean square error of the calibration model with different sub-regions,the optimum spectral range of the model was determined to be 7499.5-9403.2 cm^-1.By comparing RMSECV values,the best pretreatment method of solvent content model in cutting links was determined to be first derivative.The best principal factor of alcohol ether solvent content model was determined to be 4 by using leave-one method for mutual validation.The R_C² _V of the model is 0.9712,RMSECV is 0.396,RMSEP is 0.185.The range of prediction is about 1%,the standard deviation is less than 0.5%,and the average standard deviation is only 0.184.The average prediction deviation of a single sample is less than 0.2%.Near infrared spectroscopy?NIRS?was used to detect the pre-drying and drying exterior swing content in the flooding and dissolution control swing process.The best pretreatment method to determine the quantitative model of pre-drying external swing is multivariate scattering correction method.The best modeling interval is 9403.2-7447.2 cm^-1 and 5455.7-4598.6 cm^-1,with a principal factor of 7.The best pretreatment method for the quantitative model of pre-drying external swing is first derivative+SNV,and the best modeling interval is7507.7-5447.4 cm^-1,with the optimum principal factor is 4.The cross validation correlation coefficients R_C² _V were 0.9907 and 0.9992,the RMSECV values of cross validation root mean square error were 0.17 and 0.028,and the relative prediction deviation RPD values were 10.4and 36.4,respectively.RMSEP values of RMSEP were 0.283 and 0.0378 respectively.The average prediction deviation of samples was 0.154%and 0.028%,the average relative deviation of prediction was 1.325%and 1.389%,the average range of prediction was 0.24%and 0.183%,and the average standard deviation of prediction was 0.098%and 0.065%,respectively.The average deviation of 10 predictions for each validation set sample is less than 0.1%.