Randomfrog preferred mulberry leaf powder MicroNIR1700 spectral data modeling the best. The mean square error of calibration set, cross validation and prediction set were 1.41%, 2.34% and 2.14%, respectively, and the correlation coefficients were 0.849, 0.608 and 0.735 respectively, and Bias was -0.63%. The best results of the UVE of the DLP NIRScan Nano spectrum of the mulberry leaf   powder 

were the best. The RMSEC and R2 respectively. The RMSECV and R2 respectively.  The  RMSEP  and  R 2 of the calibration set were 1.14% and 0.921, of the cross validation were 2.13% and 0.732, of  the  prediction  set  were  2.58%  and  0.560 respectively, and Bias was -1.18%. The results show that the moisture content of mulberry leaves and the crude protein can be detected by the near infrared spectroscopy combined with wavelength optimization. At the same time, it is necessary to continue to study how to establish a prediction model for the detection of soluble sugar in mulberry leaves.

Key words: Mulberry leaf; Near Infrared Reflectance Spectroscopy; Partial Least Square Method; Wavelength selection.

目 录

第一章绪 论 1

1.1引言 1

1.2桑叶品质评价的指标 2

1.2.1粗蛋白 2

1.2.2可溶性糖 2

1.2.3含水率 3

1.3近红外光谱分析技术 3

1.3.1近红外光谱简介及发展历程 3

1.3.2近红外光谱分析技术的理论基础 4

1.3.3近红外光谱分析技术的特点 4

1.3.4近红外光谱分析技术的基本流程 5

1.3.5近红外光谱分析技术的应用 5

1.4光谱数据处理及建模 6

1.4.1偏最小二乘法