摘要
为建立阴香Cinnamomum burmannii叶片中精油右旋龙脑含量的近红外光谱预测模型,使用波通(PERTEN)公司的DA7250近红外光谱分析仪,在950~1650 nm的光谱范围内,分析了76个阴香叶片样本的光谱数据,经过光谱预处理,并比较选择最佳预处理方法、最佳光谱波段范围和最佳主成分数,采用偏最小二乘法(PLS)建立阴香叶片精油中右旋龙脑含量近红外光谱模型。结果表明:采用一阶导数-标准正态变量转换法(FD-SNV)对光谱进行预处理且当最佳主成分数为16时,得到最优模型,其校正集均方根误差(RMSEC)为7.4071,校正集相关系数(R_(C))为0.9314,交互验证集均方根误差(RMSEV)为13.4822,交互验证集相关系数(R_(V))为0.7759。说明预测值与测量值具有显著的相关性,该预测模型具有一定的准确性,可以用于阴香叶片精油中右旋龙脑含量进行快速预测。
In order to establishment of a near infrared detection model for D-borneol content in the essential oil for leaves of Cinnamomum burmannii,the spectral data of 76 leaves of C.burmannii by the DA7250 Near Infrared Spectrum Analyzer of PERTEN,Co.Ltd,in the spectral of 950 nm to 1650 nm,establishment on a near infrared detection model for D-borneol content in essential oil from leaves of C.burmannii by the PLS,after spectral preprocessing,compare and choose the best of preprocessing method,spectral band range and principal component.In a result,you can get the optimal model after preprocess the spectrum by the FD-SNV and when the principal component is 16,R_(C) is 0.9314,RMSEC is 7.4071,R_(V) is 0.7759,RMSEV is 13.4822.It directions that the predicted value has a significant correlation with the measured value,and the model has a certain prediction accuracy,it can help us to predict the D-borneol content in essential oil from leaves of C.burmannii faster.
作者
李兵
伍观娣
连辉明
林涛
姚燕飞
李毅
詹金婵
何波祥
LI Bing;WU Guandi;LIAN Huiming;LIN Tao;YAO Yanfei;LI Yi;ZHAN Jinchan;HE Boxiang(Guangdong Province Key Laboratory of Silvicuture,Protection and Utilization/Guangdong Academy of Forestry,Guangzhou,Guangdong 510520,China;Pingyuan Research Institute of Forestry,Meizhou,Guangdong 514600,China;Guangdong Huaqingyuan Biotechnology Co.,Ltd,Meizhou,Guangdong 514600,China;Guangdong Senlin Greening Co.,Ltd,Guangzhou,Guangdong 510520,China)
出处
《林业与环境科学》
2023年第1期22-29,共8页
Forestry and Environmental Science
基金
广东省林业科技创新项目(2020KJCX001,2022KJCX006,2018KJCX034)。