基于近红外光谱技术检测白酒基酒中乙醇含量的研究

为了快速测定白酒基酒中的乙醇含量，采用近红外光谱（NIRS）技术结合偏最小二乘法（PLS）处理白酒基酒近红外图谱，建 立白酒基酒中乙醇的定量分析模型。 结果表明：最佳预处理方法为减去一条直线法，最优波段谱为7 502～5 450 cm-1，主成分数为7。 校正集样品的化学值与预测值的决定系数（R2）为0.923 3，交互验证均方根误差（RMSECV）为0.89；对模型进行检验，得到验证集的 决定系数（R2）为0.954 8，预测均方根误差（RMSEP）为1.21。 结果显示近红外检测方法快速有效，所建模型具有很好的预测效果，模型 的精密度和稳定性良好。

Determination of ethanol in base liquor of Baijiu based on near infrared spectroscopy technology

In order to quickly determine the ethanol content in base liquor of Baijiu (Chinese liquor), the quantitative analysis model of ethanol in base liquor was established by using near infrared spectroscopy (NIRS) combined with partial least squares (PLS) method to deal with the near infrared spectrum of base liquor. The results showed that the optimal preprocessing method was straight line subtraction, the optimal band spectrum was 7 502- 5 450 cm-1, the principal component dimension was 7. The determination coefficient (R2) of the chemical value and the predicted value of calibration samples was 0.923 3, root mean square error of cross-validation (RMSECV) was 0.89; the model was further validated, the determinant coefficient of the validation set (R2) was 0.954 8, root mean square error of prediction (RMSEP) was 1.21. The results showed that near infrared detection method was fast and effective. The prediction effect and the precision and stability of the model was good.