摘要：
The experiment aims to establish a near-infrared fast-prediction model which measures the nutrition components of rapeseed straw using near-infrared reflectance spectroscopy (NIRS). 125 rape straw samples were collected from Gansu Province, Qinghai Province and Ningxia Hui Autonomous Region, and then the dry matter (DM), crude protein (CP), neutral detergent fiber (NDF), acid detergent fiber (ADF), ether extract (EE), coarse ash (Ash), neutral detergent insoluble protein(NDIP), acidic detergent insoluble protein (ADIP), acid detergent lignin (ADL), non-protein nitrogen (NPN) and soluble protein (SP) of the samples were determined. 100 rape straw samples were selected as calibration set to build calibration models. Another 25 samples were used as external validation set to evaluate the NIRS prediction model. The results showed that: 1) the contents of DM, CP, NDF, ADF, EE, Ash, NDIP, ADIP, ADL and NPN in rape straw were 93.62％, 5.85％,67.22％, 55.40％, 2.57％, 7.66％ and 1.04％, 0.63％, 13.58％, 1.70％, respectively. 2) The coefficient of determination for cross-validation (1-VR) and coefficient of determination for calibration of DM, CP, NDF, EE, Ash, NPN and SP were greater than 0.9 and 0.84, respectively, indicating that these NIRS models could be used in routine analysis. 3)The model parameters of DM, CP, NDF, EE, Ash, NPN and SP were Standard MSC 1,4,4,1, Weighted MSC 2,4,4,1, SNV 2,4,4,1, Standard MSC 1,4,4,1, SNV and Detrend 1,4,4,1, Weighted MSC 1,4,4,1, respectively, whereas the model of other nutritional components is not ideal and needs further improvement. In summary, the availability of nutritional composition correction model for rape straw was verified by internal and external validation.%本研究旨在利用近红外光谱技术(near infrared reflectance spectroscopy, NIRS) 建立油菜(Brassica napus) 秸秆常规成分的近红外快速预测模型.从甘肃省、青海省与宁夏回族自治区共采集125 份油菜秸秆,测定其干物质(dry matter, DM)、粗蛋白(crude protein, CP)、中性洗涤纤维(neutral detergent fiber, NDF)、酸性洗涤纤维(acid detergent fiber, ADF)、粗脂肪(ether extract, EE)、粗灰分(Ash)、中性洗涤不溶蛋白质(neutral detergent insoluble protein, NDIP)、酸性洗涤不溶蛋白质(acid detergent insoluble protein, ADIP)、酸性洗涤木质素(acid detergent lignin, ADL)、非蛋白氮(non-protein nitrogen, NPN) 与可溶性蛋白(soluble protein, SP).选取100 份油菜秸秆样品作为定标集,另外25 份作为验证集来评价NIRS 预测模型.结果显示,1) 油菜秸秆DM、CP、NDF、ADF、EE、Ash、NDIP、ADIP、ADL、NPN 和SP 的含量分别为93.62％、5.85％、67.22％、55.40％、2.57％、7.66％、1.04％、0.63％、13.58％、1.70％ 和2.89％.2)DM、CP、NDF、EE、Ash、 NPN 和SP 的交互验证决定系数(1-VR) ＞ 0.9,外部验证决定系数(RQS ≥ 0.84),构建模型可以用于日常分析.3)DM、 CP、NDF、EE、Ash、NPN 和SP 的模型参数分别为Standard MSC 1,4,4,1、Weighted MSC 2,4,4,1、SNV 2,4,4,1、Standard MSC 1,4,4,1、SNV and Detrend 1,4,4,1、Weighted MSC 1,4,4,1,其余营养成分构建模型不太理想,需进一步完善.综上,通过交互验证与外部验证验证了油菜秸秆各营养成分校正模型的可利用性.
