基于近红外技术的落叶松木材密度预测模型  

Modeling of Dahurian Larch Wood Density Based on NIR and Multivariate Analysis

作　　者：李耀翔[1] 张鸿富[1] 张亚朝[1] 张慧娟[1] 李湃[1] 

机构地区：[1]黑龙江省森林持续经营与环境微生物工程重点实验室(东北林业大学),哈尔滨150040

出　　处：《东北林业大学学报》2010年第9期27-30,共4页Journal of Northeast Forestry University

基　　金：教育部博士点基金新教师项目(200802251011);黑龙江省青年基金(QC07C59);中央高校基本科研业务费专项资金项目(DL09CB06);国家资助博士后经费

摘　　要：运用近红外光谱对落叶松(Larix gemelinii Rupr)样品密度进行了研究,分别运用偏最小二乘法及主成分回归建立预测模型,并用建立的模型分别对每一个样品进行了预测。基于偏最小二乘法的校正模型及验证模型相关系数分别为0.964和0.918,校正标准误差及预测标准误差分别为0.016和0.021,模型预测值与实测值决定系数为0.93;主成分回归模型中,校正模型及验证模型相关系数分别为0.954和0.911,校正标准误差及预测标准误差分别为0.017和0.023,模型预测值与实测值决定系数为0.91。研究表明:基于主成分回归法与偏最小二乘法的近红外光谱分析建模,都可以实现对落叶松木材密度的有效预测,但相比较而言,偏最小二乘法略优于主成分回归法,所建立的模型对落叶松木材密度预测更加准确可靠。The aim of this study was to analyse the density of Dahurian larch （Larix gmelinii Rupr.） wood samples using near-infrared （NIR） spectroscopy.Two multivariate analysis （MVA） methods of partial least squares （PLS） and principal component regression （PCR） were applied to the NIR spectroscopy of the samples.For the PLS method,the correlation coefficient （R） was 0.964 and 0.918 for the calibration and validation model,respectively.The standard error of calibration and the standard error of prediction was 0.016 and 0.021,respectively.The coefficient of determination between the predicted and measured values was 0.93.Using PCR method,the correlation coefficients were 0.954 for the calibration model and 0.911 for the validation model with standard errors of calibration and prediction of 0.017 and 0.023,respectively.The coefficient of determination between the predicted and measured values was 0.91 for PCR.Results showed that the developed models based on both PLS and PCR methods are valid to the wood samples studied.However,PLS is superior to PCR in this study in terms of model efficiency and accuracy.

关 键 词：近红外光谱 落叶松木材密度 主成分回归法 偏最小二乘法 

分 类 号：S781.31[农业科学—木材科学与技术]