LIBS结合机器学习算法的江西名优春茶采收期鉴别  

Using Laser Induced Breakdown Spectroscopy and Machine Learning to Identify Jiangxi Spring Tea Harvesting Periods

作　　者：陶雷 蔡广源 程占东 黄林[3] 何秀文[1] 徐将[1,2] 姚明印 Tao Lei;Cai Guangyuan;Cheng Zhandong;Huang Lin;He Xiuwen;Xu Jiang;Yao Mingyin(College of Engineering,Jiangxi Agricultural University,Nanchang 330045,Jiangxi,China;Key Laboratory of Biological Optoelectronics and Application in Jiangxi Universities,Nanchang 330045,Jiangxi,China;College of Bioscience and Bioengineering,Jiangxi Agricultural University,Nanchang 330045,Jiangxi,China)

机构地区：[1]江西农业大学工学院,江西南昌330045 [2]江西省高校生物光电及应用重点实验室,江西南昌330045 [3]江西农业大学生物科学与工程学院,江西南昌330045

出　　处：《激光与光电子学进展》2024年第9期479-485,共7页Laser & Optoelectronics Progress

基　　金：国家自然科学基金(32260626);国家重点研发计划子课题(2022YFD160060102);江西省教育厅科技基金项目(GJJ190189)。

摘　　要：春茶采收期极大地影响其经济价值和消费者接受度。为快速鉴别不同采收期春茶,以庐山云雾茶和狗牯脑茶的明前茶、雨前茶为对象,研究激光诱导击穿光谱(LIBS)结合机器学习的茶叶鉴别方法。每类茶叶和茶水采集100幅光谱数据,并以3∶2的比例随机划分训练集和测试集。对LIBS光谱进行基线校正预处理后优选出11组谱线数据,分别输入线性判别分析(LDA)、支持向量机(SVM)、K最近邻(KNN)、集成学习(EML)分类模型进行分析。结果表明,将茶叶茶水数据融合可有效鉴别春茶采收期,且数据融合后表现出更好的稳定性和鲁棒性,其中,LDA模型在庐山云雾春茶和狗牯脑春茶的测试集识别率分别达到98.60%、99.38%。即LIBS结合机器学习算法区分不同采收期春茶具有可行性。The harvesting period of spring tea significantly affects its economic value and consumer preference.To quickly identify different harvesting periods of spring tea,we employed laserinduced breakdown spectroscopy(LIBS)combined with machine learning algorithms.This approach was used to identify the beforebrightness tea and beforerain tea of Mt.Lushan fog tea and Dog bull head tea.One hundred spectra were collected for each type of tea leaves and tea infusion,and the training and test sets were randomly divided in a ratio of 3∶2.The LIBS spectra were preprocessed with baseline correction and then 11 sets of spectral data were preferentially selected,and input into the linear discriminant analysis(LDA),support vector machines(SVM),Knearest neighbor(KNN)and ensemble machine learning(EML)classification models for analysis,respectively.Findings showed that combining tea leaves and tea infusion data effectively identified the spring tea’s harvesting period.This fusion approach exhibited superior stability and robustness.Specifically,the LDA model achieved recognition rates of 98.60%and 99.38%in the test sets for Mt.Lushan fog tea and Dog bull head tea,respectively.Therefore,this study demonstrates the feasibility of integrating LIBS with machine learning algorithms to discern different harvesting periods of spring tea.

关 键 词：光谱学 激光诱导击穿光谱 机器学习 线性判别分析 茶叶鉴别 

分 类 号：O433.4[机械工程—光学工程]