采用近红外光谱法快速测定了单基发射药中二苯胺（DPA ）的含量。评价了多个光谱预处理方法的优化效果，采用反向偏最小二乘法（biPLS ）优选了建模波段（4698～4991、6464～6761和9414～9708 cm －1），并根据光谱分析结果和回归系数图解释了波段的选择。基于建模参数优化结果建立了二苯胺定量模型，并对模型进行了重复性验证。研究了环境温度对模型预测结果的影响。结果表明，确定的最佳光谱预处理方法是标准正态变换（SNV）、一阶导数和平滑的组合；模型校正相关系数（Rc ）和验证相关系数（Rp ）分别为0．9953和0．9885，预测均方根误差RMSEP为0．1635；重复性极差和标准偏差均低于0．1％；环境温度的变化对模型预测的影响不显著。该方法可用于快速检测单基发射药中的二苯胺含量。%Near infrared spectroscopy was applied for fast determination of diphenylamine (DPA ) content in single‐base gun propellant .The optimization effect of the multi spectral pretreatment method was evaluated .The modeling wavenumber range of 4698 -4991 ,6464 -6761 and 9414 -9708cm-1 was chosen by backward partial least squares (biPLS) and the choice of the wavenumber range was explained according to spectral analysis results and plot of regression coefficient .Based on the optimization results of the modeling parameters ,the quantitative model of diphenylamine was established and verified by repeated test .The influence of environment temperature on the model prediction was investigated .The results show that the best spectral pretreatment method determined was the combination of standard normal transformation ,first derivative and smoothing ,the correlation coefficient of calibra‐tion (Rc) and validation (Rp) of the model is 0 .9953 and 0 .9885 ,respectively and the root mean square error of predication (RMSEP) is 0 .1635 .The repeatability range and standard deviation are lower than 0 .1 % and the change in environment temperature has little influence on model predictions .The method can be used for fast deter‐mination of DPA content in single‐base gun propellant .
展开▼
