一种水溶液中果糖含量的近红外光谱分析法

A near-infrared spectroscopy analysis method for the detection of fructose content in aqueous solution

中文关键词:  近红外光谱仪  漫透射  果糖

英文关键词:near-infrared spectroscopy analysis  diffuse transmission  fructose

中文摘要:

      目的 建立一种水溶液中果糖含量的近红外光谱分析仪(near-infrared spectroscopy analysis, NIRSA)检测方法。方法 实验中所用的样品数量为30个, 随机选取26个作为校正集样品, 用于建立果糖的校正模型; 4个作为验证集样品, 用于校正模型的验证。将待测样品放入样品杯中, 利用GSA201型近红外光谱仪采集样品的光谱。得到样品的化验值, 将光谱数据和化验值导入到NIRSA化学计量学软件, 经过一阶微分S-G平滑处理, 利用成分的含量数据和光谱数据一一对应, 创建校正模型。为验证模型的预测能力, 选取4个果糖水溶液样品作为验证, 调用校正模型对该样品进行预测。结果 本研究中样品的含量范围在0.4171%~0.4431%之间, 样品的含量比较低且范围较窄, GSA近红外光谱仪测量的绝对偏差为0.000024。结论 GSA近红外光谱仪利用光谱数据和校正模型完全能够有效检测水溶液中果糖的含量, 且预测的准确度较高。

英文摘要:

      Objective To establish a method for detecting fructose content in aqueous solution by near infrared spectroscopy analysis (NIRSA). Methods The number of samples used in the experiment was 30, and 26 samples were randomly selected as calibration set samples to establish a calibration model for fructose; and 4 were used as validation set samples to verify the validation of the model. The sample to be tested was placed in a sample cup, and the spectrum of the sample was collected using a GSA201 near-infrared spectrometer. The test value of the sample was obtained, and the spectral data and the test value were imported into the NIRSA chemo metrics software, and the calibration model was created by first-order differential 9-point smoothing processing, using the content data of the component and the spectral data in a one-to-one correspondence. In order to verify the predictive ability of the model, 4 samples of aqueous fructose solution were selected for verification, and the calibration model was called to predict the sample. Results The content of the sample in this experiment ranged from 0.4171% to 0.4431%. The content of the sample was relatively low and the range was narrow. The absolute deviation measured by the GSA near-infrared spectrometer was 0.000024. Conclusion GSA NIR spectrometer can effectively detect the content of fructose in aqueous solution by using spectral data and calibration model, and the prediction accuracy is also very high.