近红外技术对南疆冬枣糖度无损检测的研究

Research on the Non-destructive Testing Technology of Near-infrared in Sugar Content of Dongzao Jujube in South of Xinjiang

筛选具有代表性的100个新疆南疆和田冬枣作为试验样品,采用近红外光谱分析技术对冬枣样品的糖度进行分析。结果表明：通过相关系数法选取特征波长区间的波数分别为5 303～5 600 cm^-1、6 205～6 302cm^-1、7 308～7 405 cm^-1,然后作多元散射校正、一阶导数和Savitzky-Golay卷积平滑法预处理后,建立的偏最小二乘法（PLS）校正模型效果最好,校正标准偏差（SEC）为0.613,预测标准偏差（SEP）为0.573,相关系数（R）为0.967,具有良好的预测效果,满足生产实践的要求。试验表明,近红外光谱分析技术适合南疆冬枣的糖度无损检测。

The aim of the study was to establish model for predicting sugar concentration of Nanjang Hetian Dongzao jujubes.One hundred Hetian Dongzao jujubes were assigned to establish model of jujubes.The sugar degree of jujubes were quantitatively determined by Near Infrared Spectroscopy（NIRS）,spectroscopy pretreatment methods were multiplication scatter correction（MSC）,first derivative,Svitzky-Golay（S-G） and Normalization,the results showed that the best effect of the calibration model of saccharinity within the spectral region of 5303-5600cm^-1,6205-6302cm^-1 and 7308-7405cm^-1 which selected through the method of correlation respectively,the model was established by Partial Least Squares,and SEC and SEP are 0.613 and 0.573,respectively,calibration coefficient and prediction correlation coefficient are 0.967 and 0.965,respectively.The final results indicated that the model is an effective method to rapid detect sugar degree.