MAPO中水分含量的近红外光谱测定

Near infrared spectrometry used for determining water content of MAPO

应用标准加入法及多种化学计量学方法研究了三-1-（2-甲基氮丙啶）氧化膦（MAPO）中水分含量的近红外光谱测定方法。分别用单波长回归、多元线性回归、主成分回归、偏最小二乘回归和人工神经网络等方法建立校正模型，用预测残差平方和评价计算结果的准确性。计算结果表明，单波长校正模型可得到更加准确的预测结果。实验结果表明，标准加入法的回收率为95％～108％，相对标准偏差为1．6％，测定结果与国家军用标准方法的测定结果无系统误差。

A near infrared spectrometry used for determining water content of tris-[ 1-（2-methyl） azifidinyl ] phosphine oxide （MAPO） was investigated by means of standard addition method and multiple stoichiometry. The calibration models were established by using single wavelength regression, multiple linear regression, principal component regression, partial least square regression and artificial neural network. The accuracy of the calculated results was estimated with predictive residual error sum of squares （PRESS）. The calculated results show that the more accurate result can be obtained by means of the single wavelength calibration model. The experimental results show that the recovery rate of standard addition method is up to 95% - 108% and the relative standard deviation （RSD） reaches 1.6%. There was no system error for the analytical results with standard addition method and national military standard method.