【作者】 王志强；

【导师】 邓国栋；

【作者基本信息】 南京理工大学 ， 材料学， 2018， 硕士

【摘要】 单基发射药是以硝化棉为主体的溶塑型火药,主要用作我国多种轻武器和火炮的发射能源,其主要工艺步骤依次为驱水、胶化、挤出成型、切断、预烘、钝感光泽、一二次烘干等。硝化棉酒精驱水是单基发射药生产工艺的第一步,酒水含量的偏差会对后续工艺产生影响,胶化过程中安定剂含量的准确性是单基发射药安定性能的基础,樟脑含量的多少影响单基发射药的钝感程度和弹道性能。对于单基发射药组分含量的检测,现阶段工厂里采用人工现场取样,再送到理化室进行分析,耗费大量的时间(>12h),且不能及时反馈组分含量的偏差。本文首次采用近红外光谱法对单基发射药生产过程中主要成分进行快速分析,具体研究内容如下:采用近红外光谱法快速检测驱水棉组分含量。通过对光谱分析优选最佳建模区间,比较交互验证均方根误差(RMSECV)和预测均方根误差(RMSEP)的值确定最优的光谱预处理方法为标准正态变化(SNV)+一阶导数+平滑,以偏最小二乘法建立水分、乙醇和NC的近红外模型,并对模型的预测能力进行验证,水分模型的校正相关系数Rc2和预测相关系数Rp2分别为0.9970和0.9978,乙醇模型的Rc2和Rp2分别为0.9835和0.9787,NC模型的Rc2和Rp2分别为0.9860和0.9931,预测值平均偏差均小于0.3%。采用近红外光谱法快速检测单基发射药中安定剂(二苯胺和二号中定剂)的组分含量。对比了多种光谱预处理方法的优化效果,确定最佳的光谱预处理方法为SNV+—阶导数+平滑和SNV+平滑,根据反向偏最小二乘法和光谱分析结果优选建模波段。基于建模优选的参数建立二苯胺和二号中定剂的近红外模型,并对模型的预测能力进行验证,二苯胺模型的Rc2和Rp2分别为0.9953和0.9874,二号中定剂模型的Rc2和Rp2分别为0.9823和0.9765,预测值平均偏差均小于0.1%,最后通过实验说明温度对安定剂模型的影响不显著。采用近红外光谱法快速检测钝感过程中樟脑组分含量。通过不同预处理方法对模型进行验证得到最优的方法为多元散射校正(MSC)+一阶导数+平滑,通过分析光谱图和OPUS7.8软件优选最佳的建模区间,以偏最小二乘法建立樟脑的近红外模型,并对模型的预测能力进行验证,樟脑模型的Rc2和Rp2分别为0.9723和0.9655,预测值平均偏差小于0.1%。更多还原

【Abstract】 Single-base gun propellant is one soluble plastic powder based on nitrocellulose(NC)as the main body.It is mainly used as the energy source of small arms and artillery in our country.The production steps of single-base gun propellant are water driving,plasticizing,extrusion molding,cutting off,pre-drying,desensitization,lustering,first/second drying,etc.Water driving from NC by alcohol is the first step in the production of single-base gun propellant,water and alcohol content of dehydrated NC have great influence on the molding process of single-base gun propellant.The accuracy of stabilizer content determines the stability of single-base gun propellant in gelling process.Camphor content has influence on the degree of desensitization and ballistic trajectory of single-base gun propellant.However,the traditional analytical methods require samples being collected from production line and sent to lab for analysis by operators.That annlytical procedures consume a long time(>12h)and cannot afford real-time quality control on production line.Near infrared(NIR)analysis was proposed as a rapid,green and in-process alternative to solve that problem,the feasibility of useing NIR analysis to determine the content of composition during the production procedure of single-base gun propellant was investigated and stated in this paper for the first time.The contents and results of this investigation were detailed below:NIR analysis was used in the fast detection of component content of single-base gun propellant after water driving.The pretreatment method of SNV+lst derivative +smooth was chosen by comparing different spectral preprocessing methods.Partial Least-Squares(PLS)method was applied to set up linear mathematical model of water,ethanol and NC content.Spectral region were selected to construct the models by spectral analysis.The correlation coefficient of calibration(Rc2)and validation(Rp2)of the water model was 0.9970 and 0.9978 respectively,the Rc2 and Rp2 of ethanol model was 0.9835 and 0.9787 respectively,the Rc2 and Rp2 of NC model was 0.9860 and 0.9931 respectively.The average deviation of predicted values were less than 0.3%.Near infrared spectroscopy was applied for fast determination of diphenylamine(DPA)and N,N’-Dimethyl carbanilide(C2)content in single-base gun propellant.The optimization effect of the multi spectral pretreatment method was evaluated,and SNV+1st derivative ＋smooth and SNV+ smooth were chosen for DPA and C2 repectively.The modeling wavenumber range was optimized by backward partial least squares(biPLS)and the spectral analysis results.Based on the optimization results of the modeling parameters,the quantitative models were established and verified.The Rc2 and Rp2 of DPA model were 0.9953 and 0.9874 respectively;the Rc2 and Rp2 of C2 model were 0.9823 and 0.9765 respectively.The average deviation of predicted values are less than 0.1%,and the change in evironment temperature has little influence on the model stability.NIR analysis was applied for fast determination of camphor content in desensitization process.The pretreatment method of MSC+1st derivative + smooth was chosen by comparing different spectral preprocessing methods.Spectral region were selected to construct the models by OPUS and spectral analysis.The Rc2 and Rp2 of the camphor model were 0.9723 and 0.9655 respectively,and the average deviation of predicted value was less than 0.1%.更多还原