猕猴桃糖度傅里叶变换近红外光谱无损检测

Using FT-NIR Spectra in Non-destructive Measurement of Kiwifruit Sugar Content

研究傅里叶漫反射近红外光谱技术在猕猴桃糖度无损检测中的应用。在近红外区域（4 000~12 000cm-1）利用偏最小二乘法（PLS）建立猕猴桃糖度与近红外光谱（NIR）的定量分析数学模型,并对不同果园、不同贮藏期、不同质量的样品所建立的糖度预测模型进行对比分析。结果表明,近红外漫反射光谱与猕猴桃糖度含量之间呈显著的线性相关,模型决定系数（R2）可达到93.65%,预测标准差RMSEP（Root Mean SquareError of Prediction）可达0.656;而且随着建模样品量及成分含量梯度的增大,预测模型通用性提高,而准确性降低。因此,建立近红外定量分析模型时,样品组成和性质应该覆盖分析样品的组成和性质的整个变化范围,以更好地满足实际应用的需要。

The nondestructive analysis of soluble-solids in kiwifruit by near infrared spectroscopy was discussed with the technique of fourier transform near infrared sapectroscopy（FT-NIRS）.In the range of 12 000 to 4 000 cm{-1},PLS calibration models were deveoloped and optimized for rapid assay analysis of soluble-solids in kiwifruit.Meanwhile,the calibration models were tested inwhich to the different orchard,the different storage period,and the different quality＇s sample were carried on the contrastive analysis.The result indicated that： the near-infrared diffuse refledlction spectrum and the soluble-solids content in kiwifruit assumes the remarkable linear correlation,the determinationcided coefficient R2 of the models R2 may achieve 93.65%,RMSEP（Root Mean Square Error of Prediction） to be possible to reach 0.656;Moreover increases along with the modelling sample quantity size and the ingredient content gradient,the model＇s versatility was enhanceds,but the reliability reduceds.So when establishinges the near-infrared quantitative calibration models,the sample composition and the nature should cover the analysis sample the composition and the nature entire range of variation,by to satisfyies the practical application well the need.