Acta Agrestia Sinica

Acta Agrestia Sinica ›› 2019, Vol. 27 ›› Issue (3): 661-667.DOI: 10.11733/j.issn.1007-0435.2019.03.019

Previous Articles     Next Articles

The Near Infrared Spectrum Analysis of the Main Mixed Herbage Nutrition in Natural Grassland,Shaertao Mountain,Xinjiang

ZHANG Fan-fan1, YU Xiao-xia1, DUAN Hong-wei2, YU lei1, ZHANG Wen-ju1   

  1. 1. College of Animal Science and Technology, Shihezi University, Shihezi 83200, China;
    2. Institute of engineering, China agricultural university, Beijing 100083, China
  • Received:2019-01-17 Revised:2019-05-22 Online:2019-06-15 Published:2019-07-19

新疆沙尔套山天然草地主要混合牧草营养指标近红外光谱分析

张凡凡1, 尉小霞1, 段宏伟2, 于磊1, 张文举1   

  1. 1. 石河子大学动物科技学院, 新疆 石河子 832000;
    2. 中国农业大学工学院, 北京 100083
  • 通讯作者: 于磊
  • 作者简介:张凡凡(1989-),男,博士,讲师,研究方向为草地及牧草资源开发与利用,E-mail:zhangfanfan@shzu.edu.cn
  • 基金资助:
    国家公益性（农业）行业专项“放牧牛羊营养均衡需要研究与示范”（201303062）资助

PDF

Abstract: In order to test main herbage nutrition in natural grassland rapidly and nondestructively,Shaertao mountain,Xinjiang,using near infrared spectroscopy (NIRS) to perform the modified partial least squares (MPLS). In combination with different spectral pretreatments and mathematical treatment methods such as scattering,derivative,smooth and other different spectra pretreatment and mathematical processing methods were used to establish the calibration model of 32 kinds main forage (grass meal) on crude protein (CP),neutral detergent fiber (NDF),acid detergent fiber (ADF) and crude Ash (Ash),calcium (Ca) and phosphorus (P). The results showed that the cross-examination determination coefficient (R2) of CP,NDF,ADF,Ash,Ca and P were 0.82,0.80,0.78,0.50,0.72 and 0.65,the standard error of cross validation (SECV) were 2.36,6.17,3.87,0.85,0.24 and 0.07,and relative percent deviation of cross validation (RPDCV) were 2.78,2.26,2.39,1.92,2.39 and 1.65,respectively. Finally,the correction model is validated by external validation set. The correlation is 0.66,0.73 and 0.84 respectively,and the accuracy of the model is still need to be improved. The detection model of Ash,Ca and P cannot be established by using NIRS.

Key words: NIRS, Natural grassland, Pasture, Nutrients, Nondestructive testing

摘要： 为了对新疆沙尔套山天然草地主要牧草营养指标进行快速、无损检测，本试验采用近红外光谱技术（Near infrared reflectance spectroscopy，NIRS），进行修正偏最小二乘回归法（Modified partial least squares，MPLS），结合散射处理、导数、平滑等不同的光谱预处理和数学处理方法，建立了32种主要牧草（草粉）的粗蛋白（Crude protein，CP）、中性洗涤纤维（Neutral detergent fiber，NDF）、酸性洗涤纤维（Acid detergent fiber，ADF）、粗灰分（Ash）、钙（Ca）和磷（P）的校正模型。结果表明：CP、NDF、ADF、Ash、Ca和P的交叉检验决定系数（R2）分别为0.82，0.80，0.78，0.50，0.72和0.65，交叉验证标准误差（Standard error of cross validation，SECV）分别为2.36，6.17，3.87，0.85，0.24和0.07，交叉验证相对分析误差（Relative percent deviation of cross validation，RPDCV）分别为2.78，2.26，2.39，1.92，2.39和1.65。最后结合外部验证集对各矫正模型进行验证。试验得出CP、NDF、ADF外部验证相对分析误差分别为2.67，2.20和2.28，相关性分别为0.66，0.73和0.84，其模型精确度和验证准确度还有待提高；利用近红外光谱检测技术不能建立Ash、Ca和P的检测模型。

关键词: 近红外光谱技术, 天然草地, 牧草, 营养成分, 无损检测

CLC Number: