能源高粱茎、叶中能源转化相关化学成分的近红外光谱模型构建与优化

何思洋; 李蒙; 唐朝臣; 周方圆; 谢光辉

本文已被：浏览 718次下载 1887次	码上扫一扫！
分享到：微信更多字体:加大+\|默认\|缩小-
能源高粱茎、叶中能源转化相关化学成分的近红外光谱模型构建与优化
何思洋^1,4,李蒙^1,2,唐朝臣^1,3,周方圆^1,4,谢光辉^1,4*
1.中国农业大学农学院, 北京 100193;2. 湖南农业大学生物科学技术学院, 长沙 410128;3. 广东省农业科学院作物研究所/广东省农作物遗传改良重点实验室, 广州 510640;4.国家能源非粮生物质原料研发中心, 北京 100193

摘要:

为建立能源高粱茎、叶中能源转化相关化学成分的近红外光谱快速定量分析模型，并探索模型优化方法，选取27份甜高粱和28份生物质高粱材料，采用化学分析方法测定茎、叶样品中可溶性糖、纤维素、半纤维素、木质素和灰分含量，分析2类能源高粱在成分含量上的差异以及茎、叶各成分含量间的相关性。同时，利用偏最小二乘法(PLS)对能源高粱茎、叶的近红外光谱建立能源转化相关化学成分分析模型，通过光谱一阶导和光谱点“竞争性自适应权重(CARS)”筛选等方法对模型进行优化。结果表明，甜高粱与生物质高粱在茎中可溶性糖、纤维素、半纤维素、木质素和叶可溶性糖含量等指标差异极显著(P<0.01)。在近红外光谱模型建立过程中，“一阶导-CARS”双优化处理使模型交叉验证决定系数(R²_CV)达到0.82~0.99，茎中可溶性糖、纤维素、半纤维素和木质素模型以及叶中可溶性糖、纤维素和灰分模型的交叉验证相对分析误差(RPD_CV)和预测相对分析误差(RPD_V)均>3.0，显著提升了模型预测性能。

关键词: 近红外光谱能源高粱木质纤维素偏最小二乘法模型优化

DOI：10.11841/j.issn.1007-4333.2021.12.04

分类号:

基金项目:国家能源局能源节约和科技装备司项目(科技司函［2012］32号)；中国大唐集团新能源股份有限公司资助项目

Construction and optimization of near-infrared spectroscopy model to analyze energy conversion chemical components in energy sorghum stems and leaves

HE Siyang^1,4,LI Meng^1,2,TANG Chaochen^1,3,ZHOU Fangyuan^1,4,XIE Guanghui^1,4*

1.College of Agronomy and Biotechnology, China Agricultural University, Beijing 100193, China;2.College of Bioscience and Biotechnology, Hunan Agricultural University, Changsha 410128, China;3.Crops Research Institute/Guangdong Provincial Key Laboratory of Crop Genetic Improvement, Guangdong Academy of Agricultural Science, Guangzhou 510640, China;4.National Energy R & D Center for Non-Food Biomass, China Agricultural University, Beijing 100193, China

Abstract:

In order to establish a quantitative near-infrared spectroscopy(NIRS)model for the rapid analyze of chemical components of energy conversion in stems and leaves of energy sorghum, and to explore the optimization method of the model, 27 sweet sorghum and 28 biomass sorghum varieties were selected. The contents of soluble sugar, cellulose, hemicellulose, lignin and ash in the stem and leaf samples of each variety were determined by chemical analysis methods. Meanwhile, the differences in the content of the two types of energy sorghum were analyzed and compared, and the correlation between the content of the stems and leaves was investigated. The partial least squares(PLS)was used to establish the NIRS models to analyze main chemical components of energy sorghum stems and leaves, and the model was optimized by first derivative and competitive adaptive reweighted sampling(CARS)selection. The results showed that there were significant differences in stem soluble sugar, cellulose, hemicellulose, lignin and leaf soluble sugar between sweet sorghum and biomass sorghum(P<0. 01). In the process of establishing NIR model, the coefficient of determination of cross validation(R²_CV)was 0. 82-0. 99 by “first derivative-CARS”. The ratio of performance to deviation of cross validation(RPD_CV)and ratio of performance to deviation of validation(RPD_V)of stem soluble sugar, cellulose, hemicellulose and lignin model, and leaf soluble sugar, cellulose and ash model were all more than 3. 0, which improved the prediction performance of the model significantly.

Key words: near-infrared spectroscopy energy sorghum lignocellulose partial least squares model optimization

引用本文: