基于NIR光谱的固体推进剂代用料混合特性  

Blending characteristics of solid propellant components based on NIR-spectroscopy

作　　者：程士超 蔺向阳[1] 李燕[1] 菅桂星 

机构地区：[1]南京理工大学化工学院,南京210094 [2]山西北方兴安化学工业有限公司,太原030008

出　　处：《固体火箭技术》2016年第3期373-377,共5页Journal of Solid Rocket Technology

基　　金：基础产品创新计划火炸药科研专项;江苏高校优势学科建设工程资助项目(PAPD)

摘　　要：建立了AOTF近红外光谱仪检测固体推进剂中多固体组分混合的方法,以PVB、MA、Al粉三固体组分在水中的混合为基础,扫描样品得到近红外光谱,经平滑、均值归一化、标准归一化(SNV)、基线校正和一阶、二阶导数预处理后,用偏最小二乘法(PLS)、交互验证建立3种组分的定量校正模型。结果表明,不同预处理方法对结果影响不一,其中使用SNV预处理消除了光谱散射对结果的影响,得到的各组分定量校正模型最优。3种组分中,定量校正模型决定系数均在0.95以上,以MA的模型决定系数最大,为0.991 377,对近红外光谱本身没有特殊吸收的组分Al粉,也得到了很好的定量校正模型,模型决定系数达到0.984 637。3种组分的定量校正模型对未知样品进行预测时,绝对误差均小于1%。因此,可用近红外光谱分析法对多固体组分的混合进行测试。A method that tests multiple solid components blending of solid propellant by NIR-spectroscopy was established. Based on the blending of PVB, MA and Al powder in water, NIR absorption spectrums were obtained by scanning samples.The spec- trams were pre-processed by smoothing, mean normalization, standard normal variate （SNV）, baseline offset, first and second deri- rate, then PLS and cross validation were used to build quantitative calibration models of the three components. The results show that different pre-processing methods have different influences. SNV pre-processing can eliminate the influence of spectral scattering, which may obtain the best quantitative calibration model. The determination coefficients are all more than 0.95, MA＇s is the biggest, 0.991 377.Al powder which hasn＇t special absorption of NIR spectrum also gets a good model, and its determination coefficient is up to 0.984 637.When this three models were usd to predict unknown sample, absolute errors are all less than 1%.Therefore, it is feasible to test the blending of multiple solid components by NIR-spectroseopy.

关 键 词：近红外光谱 固体推进剂 组分混合 定量校正模型 

分 类 号：V512[航空宇航科学与技术—航空宇航推进理论与工程]