近红外光谱技术定量测定杨梅汁可溶性固形物

采用近红外光谱分析技术对浙江省不同产地的杨梅汁进行了光谱测定和定量分析，通过计算样品的杠杆值、学生残差和马氏距离来判别异常样品，采用偏最小二乘法(PLS)对杨梅汁的可溶性固形物进行建模分析，选取不同的分辨率和波段范围对光谱进行有效的信息提取和分析，确定了最佳的回归因子数和用于定量分析的最优波段范围。结果显示： 杨梅汁样品中有一个为异常样品，在建模时予以剔除;用于杨梅汁可溶性固形物检测的最佳分辨率和最优波段分别是4 cm-1和4 000～12 267.46 cm-1,最佳的回归因子数是8，该PLS模型的相关系数为0.957 85，校正均方根误差(RMSEC)、预测均方根误差(RMSEP)和交互验证标准偏差(RMSECV)分别是0.431,0.925和1.07°Brix。研究表明近红外光谱检测技术能用于杨梅汁可溶性固形物的定量分析。

Quantitative Determination of Soluble Solid Content in Bayberry Juice Using Near-Infrared Spectroscopy Technique

Near-infrared transmittance spectra of bayberry juice of different varieties in Zhejiang province were obtained and a quantitative analysis was carried out. Leverage value, studentized residue and sample's Mahalanobis distance were applied to detect the outlier sample, and different wave number ranges and resolutions were chosen for partial least squares (PLS) regression to abstract spectral information effectively. The best factor, resolution and optimum wave number range were determined. Analysis results show that one sample was an outlier and deleted, and the best model gave the relative high correlation coefficient of 0.957 85, RMSEC, RMSEP and RMSECV of 0.431, 0.925 and 1.07 degrees Brix, respectively, when the best wave number range was 4 000-12 267.46 cm(-1), and the best factor and resolution were 8 and 4 cm(-1), respectively. The results indicate that it is feasible to use NIR spectroscopy technique for quantitative analysis of bayberry juice soluble solid content.