为建立南疆红枣品质近红外光谱在线校正模型, 分析了在线检测结果的主要影响因素, 对相应的参数进行了实验研究。 设置红枣不同的光谱采集条件, 如测量状态、 波段谱区、 仪器参数等, 利用近红外光谱仪和自行设计的红枣批量采集附件, 获取南疆红枣品质光谱, 通过光谱预处理和检测精度相结合的办法选择条件参数。 通过不同建模参数进行PLS光谱校正和二维相关光谱分析, 选择糖度特征光谱参数。 结果表明: 红枣糖度中心波长9 116, 9 418和10 500 cm-1, 采集分辨率16 cm-1, 扫描次数8次, 糖度相对误差8％～10％, 单粒红枣光谱数据量减少为原来的1/10, 采集时间减少了3 s左右。 通过实验参数, 压缩了数据量, 建立了初级在线校正模型, 基本实现南疆红枣品质近红外光谱在线检测。

To establish the on-line near infrared spectral correction model for the jujube quality of Southern Xinjiang, the main influence factors of online testing results were analyzed, and the corresponding parameters were studied. First of all, the collecting conditions of different jujube were set, such as measurement condition, spectral region, and the parameters of the apparatus. With near infrared spectrometer and self-designed jujube batch collecting attachment, the quality spectrum of jujube was obtained, and combining spectral preprocessing and detection precision, condition parameters were selected. Secondly, through PLS spectrum correction with different modeling parameters and two-dimensional correlation spectroscopy analysis, Brix characteristic spectral parameters were selected. The results showed that with sugar degree central wavelength 9 116, 9 418 and 10 500 cm-1, acquisition resolution 16 cm-1, and scan number 8, the sugar degree relative error was 8%～10%, the size of single grain jujube spectra was reduced to 1/10 of the original, and the time was reduced to 3 seconds. It was concluded that with the experimental parameters, the spectra were compressed, a primary online correction model was established, and the jujube quality on-line detection with near infrared spectroscopy was basically realized.