基于近红外光谱扫描技术对高粱中粗脂肪、粗纤维、粗灰分含量的测定方法研究

Research on Measurement of Crude Fat,Crude Fiber and Ash Contents in Sorghum Using Near-infrared Reflectance Spectroscopy Method

以收集的110份高粱样品为研究对象,采用GB/T 6433—2006、SN/T 0800.8—1999及GB/T6438—2007对粗脂肪、粗纤维、粗灰分含量进行测定。利用傅里叶变换近红外光谱仪采集高粱粉末样品的近红外漫反射光谱,光谱扫描范围4 000~12 800 cm^-1,分辨率16 cm^-1。根据定标集决定系数、定标集标准偏差、定标集交互验证决定系数、定标集交互验证标准误差等指标,确定最优模型。结果表明:高粱中粗脂肪、粗纤维、粗灰分的近红外扫描光谱交互验证集相对分析误差分别为3.93、1.47与2.37;验证集相对分析误差分别为2.57、1.23与2.34。粗脂肪的预测模型可用于实际应用,而粗纤维和粗灰分预测模型需要进一步完善。

110 sorghum samples were collected in the experiment.The chemical content of crude fat,crude fiber and ash were determined by GB/T 6433—2006,SN/T 0800.8—1999 and GB/T 6438—2007.Near infrared diffuse reflectance spectra of samples were collected by fourier transform near infrared spectrometer.The spectral scanning ranges was from 4000 to 12800 cm^-1 with resolution of 16 cm^-1.The optimal model was determined according to calibration set determination coefficient,calibration set standard deviation,calibration set interactive verification determination coefficientand,and calibration set interactive verification standard error.The results showed that the interactive validation relative analysis errorof the prediction model of crude fat,crude fiber and ash were 3.93,1.47 and 2.37 respectively,and the external validation relative standard deviation were 2.57,1.23 and 2.34 respectively.The model of crude fat was available for practical application.The models of crude fiber and ash needed to be further improved.