Study On The Optimization Model Of Near Infrared Analysis Of South Xinjiang Jujubes Sugar Degrees

South Xinjiang located in semi-arid region with resources-rich in light and heat, large temperaturedifference between day and night, provided climate conditions which are suitable for high sugar content ofred jujubes. The size and weight of jujubes are the mainly indicators for grading, while the tastequality—sugar content does not manifest. The near infrared spectroscopy (NIRS) technique with nonedestruction was applied to detect sugar degree in red jujubes which was meaningful for the brand andindustry of jujubes. The model of jujube sugar content was established based on the region and updated fortime for improving the accuracy and stability. This paper combined NIRS and Chemometrics for modeling.Firstly, study near infrared correlation spectroscopy experiment method of jujubes of South Xinjiang. Then,the modeling sample K-S method is used in the calibration set and prediction set. At last, using differentspectral preprocessing and modeling methods, near infrared spectroscopy standard model to build model bydifferent band selection method and feature selection and feature set in the spectral region of red jujubearea combined spectrum for establishing and optimizing the model.The results show that the calibration set and the division method samples expressed good modelrepresentation, objectivity and stability; and wave number range of sugar degree by the method of siPLSselected spectral region in southern red jujubes interval were6105～6402cm-1、6406～6703cm-1、7008～7305cm-1and7910～8207cm-1. At the same time the characteristic spectral region optimized by siPLSprovided a basis for designing of filter type or laser. The effect of model sugar content was best pretreatedby MSN and S-G7point3polynomial with SEC=2.06, SEP=1.92, R=0.963. the accuracy and stability ofmodel improved and valuable.