基于FT-MIR的三七粉末掺伪成分的定量检测模型研究  

Research of Quantitative Detection Model of Adulteration Components in Panax notoginseng Powder Based on FT-MIR

作　　者：何凤芸 杨晓东 周胜灵[1] 谢林 李光林[1] HE Fengyun;YANG Xiaodong;ZHOU Shengling;XIE Lin;LI Guanglin(College of Engineering and Technology,Southwest University,Chongqing 400715,China)

机构地区：[1]西南大学工程技术学院,重庆400715

出　　处：《西南大学学报（自然科学版）》2022年第4期54-61,共8页Journal of Southwest University(Natural Science Edition)

基　　金：国家自然科学基金项目(62005227).

摘　　要：高品质三七价格昂贵,常以粉末形式出售.为实现快速检测市场上三七粉末掺伪成分的定量检测,该文使用傅里叶中红外光谱技术(FT-MIR)采集三七粉末及其伪品的光谱信息,使用标准正则变换(SNV)和基线校准(Baseline)方法对原始光谱进行预处理,通过协同区间(Si)、连续投影算法(SPA)、竞争自适应加权算法(CARS)等方法选择特征变量,分别建立偏最小二乘回归(PLSR)、支持向量回归(SVR)模型.严选出能体现纯净三七粉末样品与掺杂三七粉末样品之间区别的特征变量来建立定量模型,可以提高模型鉴别精确度和稳健性.决定系数R^(2)、均方根误差(RMSE)用于评价定量模型的预测能力.实验结果表明,FT-MIR是定量检测三七粉末掺入伪品的有效方法.实验采用Si、CARS和SPA 3种变量选择方法来提高模型的稳定性和预测精度,其中基于CARS方法挑选的特征变量训练的SVR和PLSR模型对三七粉成分检测都有较好的预测效果.利用模型能够快速准确地检测三七粉末中掺入的伪品含量,对市售三七粉质量分级以及维护消费者权益、保护其生命健康方面具有较高的应用价值.High-quality Panax notoginseng(P.notoginseng)is expensive and often sold in powder form.In order to quickly detect the adulterated components of P.notoginseng powder in the market,Fourier transform mid infrared spectroscopy(FT-MIR)was used to collect the spectral information of P.notoginseng powder and its adulterants.Standard normal variate(SNV)and baseline calibration were used to preprocess the original spectrum in this work.Then characteristic variables were selected by Synergy interval(Si),successive projection algorithm(SPA),competitive adaptive weighting sampling(CARS)and so on.Partial least squares regression(PLSR)and support vector regression(SVR)models were established,respectively.Selecting the characteristic variables which can reflect the difference between pure and doped samples of P.notoginseng powder to establish the quantitative model can improve the identification accuracy and the robustness of the model to the greatest extent.The root mean square error RMSE and absolute coefficient R^(2) were used to evaluate the prediction ability of the quantitative model.The experimental results show that FT-MIR is an effective method to quantitatively detect the adulteration of P.notoginseng powder.In the experiment,three variable selection methods of Si,CARS and SPA are used to improve the stability and prediction accuracy of the model.The SVR and PLSR models trained based on the characteristic variables selected by CARS method have a great prediction effect on the composition detection of P.notoginseng powder.The model can quickly and accurately detect the content of counterfeit products in P.notoginseng powder.It has high application value for the quality classification of commercial P.notoginseng powder,safeguarding the rights and interests of consumers and protecting their life and health.

关 键 词：傅利叶中红外光谱技术 三七粉末 定量检测 支持向量回归 偏最小二乘回归 

分 类 号：S567.236[农业科学—中草药栽培]