兰州大学机构库 >草地农业科技学院
利用NIRS技术预测秸秆及苜蓿干草营养组成的研究
Alternative TitleEvaluation of Nutritional Composition of Straw and Alfalfa Hay by Using NIRS Technology
郭涛
Subtype硕士
Thesis Advisor李飞
2020-07-31
Degree Grantor兰州大学
Place of Conferral兰州
Degree Name农学硕士
Degree Discipline畜牧学
Keyword玉米秸秆 小麦秸秆 苜蓿干草 营养成分 近红外预测模型
Abstract本研究旨在利用近红外光谱技术分别建立玉米秸秆、小麦秸秆和苜蓿干草近红外预测模型,并且比较苜蓿干草不同处理方式对于建立的近红外预测模型的预测准确性的影响。本论文包括以下3个试验: 试验1:本研究旨在利用近红外光谱技术(near-infrared reflectance spectroscopy, NIRS)分别建立玉米秸秆(corn straw)和小麦秸秆(wheat straw)的近红外预测模型。从甘肃、新疆和河南三个省份共采集玉米秸秆样品155份,小麦秸秆样品135份。选取玉米秸秆124份作为定标集,31份作为验证集。选取小麦秸秆108份作为定标集,27份作为验证集。利用近红外光谱技术结合改良偏最小二乘法(Modified partial least squares, MPLS)等化学计量学方法分别建立玉米秸秆和小麦秸秆的干物质(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)和酸性洗涤木质素(acid detergent lignin, ADL)这5个指标的近红外预测模型。结果表明,1)玉米秸秆平均DM、CP、NDF、ADF和ADL的含量分别为94.60%、5.16%、63.88%、36.33%和3.32%。小麦秸秆平均DM、CP、NDF、ADF和ADL的含量分别为95.35%、3.42%、77.31%、46.59%和6.84%。2)玉米和小麦秸秆的CP含量的预测模型交互验证决定系数(1-VR)>0.90,且外部验证决定系数(RSQ)>0.84,构建的模型可以用于实际预测。3)玉米秸秆DM、NDF、ADF和小麦秸秆DM各指标定标模型的1-VR值在0.80左右,可以粗略的预测其营养成分含量,其余各指标模型预测效果不太理想,模型需要进一步优化。综上所述,本研究为生产实践中快速预测玉米和小麦秸秆营养成分含量提供了理论依据,并且通过NIRS建立了其近红外预测模型。 试验2:本试验旨在利用NIRS建立苜蓿干草(Alfalfa hay) 6种营养成分的近红外预测模型。分别从甘肃、宁夏、河北、江苏和陕西5个省份采集200份苜蓿干草样品,测定其DM、粗灰分(Ash)、CP、NDF、ADF和粗脂肪(ether extract, EE)的含量。选取苜蓿干草样品160份作定标集,40份作验证集。利用NIRS结合MPLS构建并且验证其建立预测模型的优劣。结果表明:苜蓿干草DM、NDF含量预测模型的预测决定系数(RSQ)和外部验证相对分析误差(RPD)分别为0.87和2.67,0.90和3.16,构建的模型可以用于实际生产中的预测CP、ADF含量预测模型的RSQ和RPD分别为0.83和2.41,0.82和2.28,构建的预测模型不能完全代替湿化学分析,但可以用于大量样品的筛选分析Ash含量预测模型的RSQ和RPD为0.59和1.51,构建的预测模型只能用于粗略的分析EE含量预测模型的RSQ和RPD为0.45和1.32,构建的预测模型相关性较差,还需进一步优化。 试验3:本试验旨在利用NIRS分别建立切短和粉碎苜蓿干草DM、Ash、CP、NDF、ADF和EE近红外预测模型,选取苜蓿干草样品149份作为定标集,37份作为验证集,利用改良偏最小二乘法建立各指标预测模型,分析切短和粉碎苜蓿干草各近红外预测模型的预测偏差。结果表明,DM和Ash两种处理方式预测模型预测偏差P>0.05,差异不显著。切短和粉碎CP、NDF、ADF和EE预测模型预测偏差P<0.01,差异极显著。切短苜蓿干草6种营养成分近红外预测模型预测效果较差。 综上所述,利用NIRS建立了玉米秸秆、小麦秸秆的DM、CP、NDF、ADF、ADL和苜蓿干草DM、Ash、CP、NDF、ADF、 EE各营养成分预测模型,为快速评定其营养价值提供便捷,高效合理利用这3种粗饲料资源提供理论依据。
Other AbstractThe purpose of this study was to use near infrared spectroscopy to establish near infrared prediction models of corn stalks, wheat straws and alfalfa hay, respectively, and compare the effects of different treatment methods of alfalfa hay on the prediction accuracy of the established near infrared prediction models. This paper includes the following three experiments: Experiment 1: The experiment aimed to establish a near-infrared prediction model for corn straw and wheat straw by using near-infrared spectroscopy (NIRS).A total of 155 samples of corn straw and 135 samples of wheat straw were collected from three provinces of Gansu, Xinjiang and Henan. A total of 124 corn stalks were selected as the calibration set, and 31 were used as the verification set.108 wheat straws were selected as the calibration set and 27 were used as the verification set. The near-infrared prediction models of DM, CP, NDF, ADF and ADL of corn straw and wheat straw were established by near-infrared spectroscopy combined with modified partial least squares (MPLS) and other stoichiometry methods. The results showed that 1) the average contents of DM, CP, NDF, ADF and ADL in corn straw were 94.60%, 5.16%, 63.88%, 36.33% and 3.32%, respectively. The average contents of DM, CP, NDF, ADF and ADL in wheat straw were 95.35%, 3.42%, 77.31%, 46.59% and 6.84%, respectively. 2) The prediction model of the CP content of corn and wheat straw is interactive verification coefficient (1-VR)>0.90, and the external verification determination coefficient (RSQ) is >0.84. The constructed model can be used for actual prediction. 3) The 1-VR value of the calibration model for corn stalks DM, NDF, ADF and wheat straw DM is over 0.80, which can roughly predict the nutrient content. The prediction results of other indicators are not ideal, and the model needs further optimization. In summary, the study provides a theoretical basis for the rapid prediction of nutrient content of corn and wheat straw in production practice, and established its near-infrared prediction model through NIRS. Experiment 2: The experiment aimed to establish a near-infrared prediction model for six nutrients of alfalfa hay by using near-infrared reflectance spectroscopy (NIRS). 200 samples of alfalfa hay were collected from five provinces of Gansu, Ningxia, Hebei, Jiangsu and Shanxi to analyze DM、Ash、CP,neutral detergent fiber. (NDF), acid detergent fiber (ADF) and ether extract (EE). The calibration sets and verification sets included 160 samples and 40 samples, respectively. The NIRS is combined with Modified partial least squares (MPLS) to construct and verify the prediction model which is accurate and inaccurate. The results show that the coefficient of determination for validation (RSQ) and ratio of performance to deviation for validation (RPD) of the DM and NDF prediction models of alfalfa hay were 0.87 and 2.67, 0.90 and 3.16, respectively. It was constructed that the model can be used for prediction in actual production. The prediction model RSQ and RPD of the CP and ADF content were 0.83 and 2.41, 0.82 and 2.28, respectively. The constructed prediction model can not completely replace the wet chemical analysis, but can be used for the screening analysis of a large number of samples. The ash content Zprediction model RSQ and RPD were 0.59 And 1.51. The constructed prediction model can only be used for rough analysis. the RSQ and RPD of the EE content prediction model are 0.45 and 1.32, and the constructed prediction model is poorly correlated and needs further optimization. Experiment 3: The purpose of this experiment was to use NIRS to establish the near infrared prediction models of DM, Ash, CP, NDF, ADF and EE for chopped and crushed alfalfa hay. The least square method was used to establish the prediction model of each index, and the prediction deviation of each near infrared prediction model of cut and crushed alfalfa hay was analyzed. The results show that the prediction deviations of the prediction models of DM and Ash are P >0.05, and the difference is not significant. The short-cut and crushed CP, NDF, ADF and EE prediction models had a prediction deviation P <0.01, the difference was very significant. The near infrared prediction model of 6 nutrient components of cut alfalfa hay has poor prediction effect. In summary, It was established the corn straw, wheat straw and alfalfa hay prediction models by NIRS. It was aimed to analyze their nutritional value. Provide convenient, efficient and rational use of these three coarse feed resources to provide theoretical basis.
Pages55
URL查看原文
Language中文
Document Type学位论文
Identifierhttps://ir.lzu.edu.cn/handle/262010/467328
Collection草地农业科技学院
Affiliation
草地农业科技学院
First Author AffilicationCollege of Pastoral Agriculture Science and Technology
Recommended Citation
GB/T 7714
郭涛. 利用NIRS技术预测秸秆及苜蓿干草营养组成的研究[D]. 兰州. 兰州大学,2020.
Files in This Item:
There are no files associated with this item.
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Altmetrics Score
Google Scholar
Similar articles in Google Scholar
[郭涛]'s Articles
Baidu academic
Similar articles in Baidu academic
[郭涛]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[郭涛]'s Articles
Terms of Use
No data!
Social Bookmark/Share
No comment.
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.