文献类型：期刊文献

中文题名：制浆材综纤维素和聚戊糖含量的近红外预测

英文题名：The Prediction of Holocellulose and Pentosan Content in Pulpwood by Near Infrared Spectroscopy

作者：吴珽[1] 房桂干[1] 梁龙[1] 张新民[2] 赵振义[2]

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江苏省生物质能源与材料重点实验室国家林业局林产化学工程重点开放性实验室生物质化学利用国家工程实验室;[2]华夏科创仪器有限公司

年份：2015

卷号：23

期号：4

起止页码：49-54

中文期刊名：纤维素科学与技术

外文期刊名：Journal of Cellulose Science and Technology

收录：CSCD:【CSCD2015_2016】；

基金：国家林业局948项目"农林剩余物制机械浆节能和减量技术引进"(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;综纤维素;聚戊糖

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

分类号：O657.3;TS721

摘要：用传统方法测定了156个制浆材样品的综纤维素和聚戊糖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,确定最佳主成分数分别为9和10并建立样品综纤维素和聚戊糖含量的校正模型。独立验证中两个模型的决定系数R_（val）~2分别为0.903 4、0.940 1,预测均方根误差（RMSEP）分别为0.69%、0.78%,相对分析误差（RPD）值分别为3.22、4.09,绝对偏差（AD）分别为-1.00%~1.20%、-1.39%~1.31%,两个校正模型较好地预测了验证集样品的综纤维素和聚戊糖含量,基本满足制浆造纸工业中快速测定的需求。
A study was conducted to use the traditional method to analyze the content of holocellulose and pentosan of 156 pulpwood samples and the near-infrared （NIR） spectra were also collected. After the original spectra were pretreated by multipliplicative scatter correction （MSC）, partial least squares （PLS） method and cross-validation was used to confirm the best factors would be 9 and 10. Based on those factors, calibration models for holocellulose and pentosan content was built. The independent verification of the calibration models showed the coefficient of determination （Rval2） was 0.9034 and 0.9401. The root mean square error of prediction （RMSEP） was 0.69% and 0.78%. The relative percent deviation （RPD） was 3.22 and 4.09. And the absolute deviation （AD） was -1.00%-1.20% and -1.39%-1.31%. Both calibration models basically met the requirements of rapid determination in pulping and paper making industry for their good predictive performance.