文献类型：期刊文献

中文题名：近红外光谱法快速测定制浆材化学成分含量

英文题名：The rapid prediction of chemical components contents in pulpwood by near infrared spectroscopy

作者：吴珽[1] 房桂干[1,2] 梁龙[1] 崔宏辉[1] 熊智新[2,3]

第一作者：吴珽

机构：[1]中国林业科学研究院林产化学工业研究所;[2]南京林业大学林业资源高效加工利用协同创新中心;[3]南京林业大学轻工科学与工程学院

年份：2016

卷号：1

期号：2

起止页码：77-81

中文期刊名：林业工程学报

外文期刊名：Journal of Forestry Engineering

收录：CSTPCD;;北大核心:【北大核心2014】；

基金：国家林业局"948"项目(2014-4-31)

语种：中文

中文关键词：近红外光谱;偏最小二乘法;制浆材;化学成分

外文关键词：near-infrared spectroscopy;partial least squares（PLS）;pulpwood;chemical component

分类号：O657.33;TQ351

摘要：为实现制浆材化学成分含量的快速测定与实时分析,用常规方法测定了120个制浆材样品的综纤维素、聚戊糖、酸不溶木素及苯醇抽出物含量,并采集了样品的近红外光谱。对原始光谱进行多元散射校正后,运用偏最小二乘法和交互验证的方法,根据主成分数与PRESS值的关系,确定最佳主成分数分别为9,10,8,9,据此建立样品相关化学成分含量的校正模型。独立验证中综纤维素、聚戊糖、酸不溶木素和苯醇抽出物模型的决定系数R2val分别为0.918 8,0.949 3,0.946 6和0.928 4,预测均方根误差（RMSEP）分别为0.70%,0.75%,0.72%和0.24%,相对分析误差（RPD）值分别为3.50,4.44,4.33和3.74,绝对偏差（AD）分别为-0.95%~1.22%,-1.42%~1.29%,-1.39%~1.14%和-0.34%~0.39%,4个校正模型较好地预测了验证集样品的化学成分含量,基本满足制浆造纸工业中快速测定的需求。
In order to realize rapid determination and real-time analysis of chemical components contents in pulpwood,the content of holocellulose,polypentanose,acid-insoluble lignin and benzene-alcohol extractives in 120 pulpwood samples were analyzed by means of conventional methods,and their near-infrared（ NIR） spectroscopy were collected. Based on the pretreatment of the original spectra by multipliplicative scatter correction（ MSC）,the factors were confirmed by means of partial least squares（ PLS） method and cross-validation. On the basis of the relationship between factors and PRESS value,the four optimum factors were confirmed as 9,10,8,9,respectively. According to the four optimum factors,calibration models for the content of holocellulose,polypentanose,acid-insoluble lignin and benzene-alcohol extractives were built. The independent verification showed that the coefficient of determination（ R2val） of the holocellulose,polypentanose,acid-insoluble lignin and benzene-alcohol models were 0. 918 8,0. 949 3,0. 946 6 and 0. 928 4,respectively,and the root mean square error of prediction（ RMSEP） were 0. 70%,0. 75%,0. 72% and 0. 24%,respectively. The relative percent deviation（ RPD） were 3. 50,4. 44,4. 33 and 3. 74,respectively,and the absolute deviation（ AD） were- 0. 95% to1. 22%,-1. 42% to 1. 29%,-1. 39% to 1. 14% and-0.34% to 0. 39% respectively. All the calibration models basically meet the requirements of rapid determination in pulp and paper industry with their good predictive performance.