| 摘要: |
| 基于近红外光谱技术 (Near infrared reflectance spectroscopy, NIRS), 对广东省主要五指毛桃
Ficus hirta 产地样品的水分及活性成分进行测定, 建立定量模型并进行预测效果评价, 实现五指毛桃成分
含量的快速检测。 研究在广东省湛江、 茂名、 云浮、 河源、 梅州 5 个主要产地采集五指毛桃 292 份, 以高
效液相色谱测定的质量分数作参比, 采用 NIRS 结合偏最小二乘法, 建立根中水分、 补骨脂素及佛手柑内
酯含量的快检定量模型。 结果表明, 五指毛桃水分含量、 补骨脂素及佛手柑内酯质量分数的定量模型相
关系数为 0. 97、 0. 91 和 0. 83; 校正均方根偏差为 0. 23、 0. 18、 0. 03; 交互验证均方根偏差为 0. 27、
0. 25、 0. 04; 预测均方根误差为 0. 20、 0. 20 和 0. 03。 五指毛桃 NIRS 定量模型中, 水分模型稳定性和准
确性较好, 检测精度达到近红外快速检测设备通用要求; 补骨脂素模型预测效果较好, 相关性良好; 佛
手柑内脂模型效果较差, 该指标含量较低, 且分布梯度相对集中, 建议作为初筛参考。 |
| 关键词: 五指毛桃 活性成分 近红外光谱技术 定量模型 |
| DOI： |
| 分类号: |
| 基金项目:广东省林业科技创新项目 (2021KJCX004、 2022KJCX012)。 |
|
| Models for Determining Active Ingredients Content in Ficus hirta by Using Near Infrared Spectroscopy |
|
wang song, liu lu bing, wang zhi hong, xu qiao lin, chen ying le, huang wen yan, yang liu, zeng lei
|
|
Guangdong Academy of Forestry
|
| Abstract: |
| By using near infrared reflectance spectroscopy (NIRS), the water content and active ingredients of Ficus hirta collected from Guangdong province were determined, and the corresponding quantitative
models were established, and the prediction effect of models was evaluated. The rapid determination of water
content and active ingredient content in F. hirta was established. 292 samples of F. hirta were collected from five
major producing areas of Guangdong province, including Zhanjiang, Maoming, Yunfu, Heyuan and
Meizhou. On the basis of high performance liquid chromatography (HPLC), using NIRS combined with partial least squares ( PLS), NIRS quantitative models were established for the determination of water content,
psoralen and bergapten content in the root of F. hirta. The correlation coefficients for water content, psoralen
and bergapten content were 0. 97, 0. 91 and 0. 83 respectively, and RMSEC were 0. 23, 0. 18, 0. 03 respec-tively, and RMSECV were 0. 27, 0. 25 and 0. 04 respectively, and RMSEP were respectively 0. 20, 0. 20 and
0. 03. The NIRS quantitative models of water content were efficient and robust, and the detection accuracy
reached general requirements for near infrared rapid detection equipment, which can satisfy the daily needs of
detection. The NIRS quantitative models of psoralen content have good effect and good correlation. The effect
of the bergapten content model is poor, due to the low content of this index and the relatively concentrated distribution gradient. So it is suggested that NIRS can be used as the reference value of bergapten content initial
screening in the root of F. hirta. |
| Key words: Ficus hirta active ingredients NIRS quantitative models |
