引用本文:
【打印本页】   【下载PDF全文】   查看/发表评论  下载PDF阅读器  关闭
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 280次   下载 237 本文二维码信息
码上扫一扫!
分享到: 微信 更多
湿加松16种黄酮总含量近红外预测模型的建立
司徒荣贵1, 钟岁英1, 叶威方1, 司徒文斗1, 毛积鹏2
1.台山市红岭种子园;2.台山市红岭种子园/华南农业大学林学与风景园林学院
摘要:
在950~1 650 nm 的光谱范围内,使用DA2700 型近红外光谱仪采集了110 个湿加松Pinus elliottii × P. caribaea 松针粉末样本的光谱数据。结合实际测定值,采用偏最小二乘(PLS)回归法并选择最佳光谱预处理方法和最佳主成分数,建立湿加松松针组织16 种黄酮物质总含量的快速预测模型。结果表明:采用一阶导数(FD)与滤波拟合(SG)相结合法对光谱数据进行预处理且当主成分数为10 时,可得最优模型。其校正集相关系数(Rc)和交互验证集相关系数(Rv)分别为0.852 1 和0.705 9。校正集均方根误差(RMSEC)和交互验证集均方根误差(RMSEV)分别为6.361 0 和9.150 9。外部验证集测定值和模型预测值之间的相关系数为0.853 7。综合表明所建模型预测精度高、可靠性强,可用于湿加松松针组织16 种黄酮物质总含量的预测。
关键词:  湿加松  黄酮化合物  近红外  预测模型
DOI:
分类号:
基金项目:
Establishment of Near Infrared Prediction Model for Determination of Total Content of 16 Flavonoids in Pinus elliottii × P. caribaea
si-tu rong gui1, zhong sui ying1, ye wei fang1, si-tu wen dou1, Mao ji peng2
1.Taishan Hongling Seed Orchart;2.Taishan Hongling Seed Orchart/College of Forestry and Landscape Architecture, South China Agricultural University
Abstract:
The spectral data of 110 Pinus elliottii × P. caribaea samples were collected within the wavelength of 950~1 650 nm using DA2700 near infrared spectroscopy (NIRS). Combined with the actual measured values, and the partial least squares (PLS) regression method, the optimal spectral pretreatment method and the optimal principal component number were selected to establish the rapid prediction model for the total content of 16 flavonoids. The results show that the optimal model can be obtained when First Deviation (FD) and Savitzky Golay (SG) methods are used to pretreatment the spectral data and the principal component number is 10. The related coefficient of calibration (Rc) and related coefficient of validation (Rv) were 0.852 1 and 0.705 9, respectively. The root mean square error of calibration (RMSEC) and root mean square error of validation (RMSEV) were 6.316 0 and 9.150 9, respectively. The correlation coefficient between the measured values and the predicted values of external validation set was 0.853 7. The results showed that the established model had high accuracy and strong reliability, and could be used to predict the total content of 16 flavonods in needle of P. elliottii × P. caribaea.
Key words:  Pinus elliottii × P. caribaea  flavonods  near infrared spectroscopy  prediction model