基于傅里叶变换红外光谱指纹技术的艾叶产地溯源研究

中药材产地对中药材品质与安全有直接影响。从生物学角度，中药材是物种受特定生态环境的影响，在长期生态适应过程中所形成的。药材生长所需的气候、土壤、水文等生态因素与其生长发育和品质唇齿相关，并带有地域信息的指纹特征。近年来，中医药产业的快速发展带来了中药材资源需求量的激增，但同时也存在诸多安全隐患，药材产地难以区分和溯源已成为制约中医药发展的主要瓶颈之一。以国内4个省份5个主产区的75份艾叶样品为实验材料，采用FTIR法进行红外特征分析和数据挖掘，通过比较多种光谱信号预处理方法(如高斯滤波、多元散射校正、标准正态变换、一阶/二阶导数等)和多种模式识别技术(如BP神经网络、随机森林、 K-最近邻算法、贝叶斯算法、粒子群优化支持向量机模型等),探索适合艾叶产地溯源的计量学方法。结果表明，K-最近邻算法、贝叶斯算法及粒子群优化支持向量机3种模式识别效果最为理想，测试集的正确率均为100%。基于运行时间、鉴别正确率与模型稳定性综合考虑，最终确定K-最近邻算法是艾叶产地鉴别的最优方法。整体来看，红外光谱指纹技术结合适当的化学计量学方法能够用于艾叶的产地溯源，研究结果为艾叶的道地性评价和质量控制提供技...

Study on Geographical Traceability of Artemisia argyi by Employing the Fourier Transform Infrared Spectral Fingerprinting

The geographical distribution of medicinal plants significantly affect the quality and safety of Chinese medicinal materials. From the biological point of view, Chinese medicinal materials are formed during the long-term ecological adaptation of species affected by a specific ecological environment. The climate, soil, hydrology, and other ecological factors required for the growth of medicinal materials are closely related to their growth and quality and have fingerprint characteristics of geographical information. In recent years, the rapid development of the Chinese medicine industry has brought about a surge in demand for Chinese medicine resources. However, at the same time, there are also many potential safety hazards. The difficulty in distinguishing and tracing the origin of Chinese medicinal materials has become one of the main bottlenecks restricting the development of traditional Chinese medicine. In this study, 75 A. argyi samples from 5 major producing areas of 4 provinces in China were analyzed by FTIR for characteristic analysis and data mining. Spectral signal preprocessing methods include Gaussian filtering, multivariate scattering correction, standard normal transformation, first/second derivative, etc. and pattern recognition techniques include BP neural network model, random forest, K-nearest neighbor, Bayesian algorithm, particle swarm optimization support vector machine, etc. were applied to explore the feasibility of traceability for A. argyi. The results indicate that the algorithms of K-nearest neighbor, Bayesian, and particle swarm optimization support vector machine show the ideal recognition effect, with an accuracy of 100%. Considering the comprehensive factors of running time, identification accuracy, and model stability, the algorithm of K-nearest neighbor is determined as the best method to trace the origin of A. argyi. In general, FTIR technology combined with appropriate chemometrics methods can be used to trace the origin of A. argyi successfully. The results of this study can provide technical support for the evaluation and quality control of A. argyi, and also contribute useful reference for the isotropic research of other medicinal materials.