蜂蜜中果糖含量近红外光谱试验研究

Study on the fructose content of honey by using fourier transform near-infrared spectroscopy

以68份蜂蜜样品为材料,利用傅立叶变换近红外光谱技术(NIR)结合偏最小二乘法(PLS),建立了蜂蜜中果糖含量的近红外光谱数学模型,并对模型预测结果进行评价。结果表明,选择11065.36cm-1～7857.90cm-1波数范围内的光谱,进行SG预处理方法,用8个主成分建立的蜂蜜果糖含量模型效果最好,模型相关系数、校正均方差和预测均方差分别为0.9259、1.21和1.64。该方法具有快速、准确、成本低等特点。

Mathematic models for analysis of fructose contents in 68 honey samples were established by using fourier transform near-infrared spectroscopy（FT-NIR） combined with partial least squares（PLS）, and the predicted results were also estimated. Through selecting S.G. preprocessing, 8 principal compents and 11065.36cm^-1～7857.90cm^-1spectra range, the model was suitable. The results showed that the correlation coefficient （R）, the root mean square errors of cross validation（RMSECV） and the root mean square errors of prediction（RMSEP） were 0.9259、1.21and 1.64, respectively. The method has the advantage of shorter testing time, accuray and low cost.