Determination of hemicellulose, cellulose and lignin content using visible and near infrared spectroscopy in Miscanthus sinensis
Determination of hemicellulose, cellulose and lignin content using visible and near infrared spectroscopy in Miscanthus sinensis
复制标题
DOI：
10.1016/j.biortech.2017.05.047
复制
发表时间：
2017-10-01
影响因子：
11.4
通讯作者：
Peng, Junhua
中科院分区：
文献类型：
--
作者：
Jin, Xiaoli;Chen, Xiaoling;Peng, Junhua
包括半纤维素、纤维素和木质素的木质素纤维素组分是植物细胞壁的三种主要组分，并且它们在生物质作物（例如芒草（Miscanthus sinensis））中的比例极大地影响原料转化为液体燃料或生物产品。本研究探讨了利用维斯/近红外光谱快速定量测定半纤维素、纤维素和木质素的可行性。sinensis进行了调查。首先，采用偏最小二乘（PLS）、最小二乘支持向量机回归（LSSVR）和基于全波长的径向基函数神经网络（RBF_NN）建立预测模型。随后，23，25和27个特征波长的半纤维素，纤维素和木质素，分别被发现，显示出显着的贡献，校准模型。基于特征波长分别采用PLS、LS-SVM和ANN建立了三种测定模型。成功地建立了木质纤维素组分的校正模型，并应用于M. sinensis。(C)2017爱思唯尔有限公司版权所有
Lignocellulosic components including hemicellulose, cellulose and lignin are the three major components of plant cell walls, and their proportions in biomass crops, such as Miscanthus sinensis, greatly impact feed stock conversion to liquid fuels or bio-products. In this study, the feasibility of using visible and near infrared (VIS/NIR) spectroscopy to rapidly quantify hemicellulose, cellulose and lignin in M. sinensis was investigated. Initially, prediction models were established using partial least squares (PLS), least squares support vector machine regression (LSSVR), and radial basis function neural network (RBF_NN) based on whole wavelengths. Subsequently, 23, 25 and 27 characteristic wavelengths for hemicellulose, cellulose and lignin, respectively, were found to show significant contribution to calibration models. Three determination models were eventually built by PLS, LS-SVM and ANN based on the characteristic wavelengths. Calibration models for lignocellulosic components were successfully developed, and can now be applied to assessment of lignocellulose contents in M. sinensis. (C) 2017 Elsevier Ltd. All rights reserved.