返回 
加入收藏
贯叶金丝桃质控方法提升及“辨色论质”研究
x
请在关注微信后，向客服人员索取文件
| 篇名： | 贯叶金丝桃质控方法提升及“辨色论质”研究 |
| TITLE： | Improvement of quality control methods and “quality evaluation via color discrimination”of Hypericum perforatum |
| 摘要： | 目的 为贯叶金丝桃的质量控制提供参考。方法采用高效液相色谱法建立20批贯叶金丝桃的指纹图谱并测定其主要成分绿原酸、芦丁、金丝桃苷、异槲皮苷、萹蓄苷、槲皮苷、槲皮素的含量；采用SPSS26.0软件进行聚类分析。采用电子眼测定贯叶金丝桃粉末的明度值（L*）、红绿值（a*）和黄蓝值（b*），采用机器学习算法建立基于外观色度值的贯叶金丝桃上述7种成分含量的预测模型，并采用均方根误差（RMSE）评价预测模型的预测性能。结果20批贯叶金丝桃指纹图谱共标定16个共有峰，指认出9个色谱峰，分别为绿原酸、芦丁、金丝桃苷、异槲皮苷、萹蓄苷、槲皮苷、槲皮素、贯叶金丝桃素和金丝桃素，20批样品与对照图谱的相似度为0.889～0.987；绿原酸、芦丁、金丝桃苷、异槲皮苷、萹蓄苷、槲皮苷、槲皮素含量分别为0.025％～0.166％、0.048％～0.339％、0.082％～0.419％、0.017％～0.209％、0.011％～0.134％、0.020％～0.135％、0.041％～0.235％；聚类分析结果显示，当欧氏距离为1.4时，18批合格贯叶金丝桃可聚为3类。20批贯叶金丝桃的L*为62.814～75.668，a*为1.409～3.490，b*为25.249～30.759；XGBoost、LightGBM、AdaBoost3种预测模型的RMSE为0.008～0.070，拟合效果良好。除芦丁外，XGBoost模型预测其余6种成分的含量均具有较高的预测精度。结论所建指纹图谱及含量测定方法准确、重复性好、结果可靠；结合机器学习算法构建的基于外观色度值的含量预测模型可用于贯叶金丝桃的质量控制。 |
| ABSTRACT： | OBJECTIVE To provide a reference for the quality control of Hypericum perforatum. METHODS High- performance liquid chromatography （HPLC） was used to establish fingerprints for 20 batches of H. perforatum and determine the contents of its main components： chlorogenic acid， rutin， hyperin， isoquercitrin， avicularin， quercitrin and quercetin. Cluster analysis was conducted using SPSS 26.0 software. The chromaticity values （luminance value L*， red-green value a*， and yellow- blue value b*） of H. perforatum powder were measured using electronic eye. A prediction model for the contents of seven components in H. perforatum based on its appearance chromaticity values was established using machine learning algorithms. The predictive performance of the models was evaluated using root-mean-square-error （RMSE）. RESULTS A total of 16 common peaks were calibrated in the fingerprints of 20 batches of H. perforatum， and 9 peaks were identified， which were chlorogenic acid， rutin， hyperin， isoquercitrin， avicularin， quercitrin， quercetin， hypericin and hyperforin； the similarities of the 20 batches of samples and reference fingerprint ranged from 0.889-0.987. The contents of chlorogenic acid， rutin， hyperin， isoquercitrin， avicularin， quercitrin and quercetin were 0.025%-0.166%， 0.048%-0.339%， 0.082%-0.419%， 0.017%-0.209%， 0.011%-0.134%， 0.020%-0.135%， 0.041%-0.235%， respectively. Cluster analysis results showed that 18 batches of qualified H. perforatum were grouped into three categories， when the Euclidean distance was set to 1.4. L* of the 20 batches of H. perforatum ranged from 62.814 to 75.668， a* ranged from 1.409 to 3.490， and b* ranged from 25.249 to 30.759. RMSE of three prediction models， namely XGBoost， LightGBM， and AdaBoost， ranged from 0.008 to 0.070， indicating good fitting performance. XGBoost model predicted the contents of the other six components with high accuracy， except for rutin. CONCLUSIONS The established fingerprints and content determination methods are accurate， reproducible， and reliable. The content prediction model based on appearance chromaticity values， combined with machine learning algorithms， can be used for the quality control of H. perforatum. |
| 期刊： | 2025年第36卷第06期 |
| 作者： | 李喜硕；苏本正；曲珍妮；朱娟娟；戴衍朋；石典花 |
| AUTHORS： | LI Xishuo，SU Benzheng，QU Zhenni，ZHU Juanjuan，DAI Yanpeng，SHI Dianhua |
| 关键字： | 贯叶金丝桃；指纹图谱；聚类分析；含量测定；色度值；质量评价；机器学习；预测模型 |
| KEYWORDS： | Hypericum perforatum； fingerprint； cluster analysis； content determination； chromaticity values； quality |
| 阅读数： | 3 次 |
| 本月下载数： | 0 次 |
* 注：未经本站明确许可，任何网站不得非法盗链资源下载连接及抄袭本站原创内容资源!在此感谢您的支持与合作！
