基于拉曼光谱和线性判别对于洗衣皂的检测研究  

Research on laundry soap based on Raman spectroscopy and linear discrimination

作　　者：侯赛文 李春宇[1] HOU Sai-wen;LI Chun-yu(College of Investigation,People's Public Security University of China,Beijing 100038,China)

机构地区：[1]中国人民公安大学侦查学院,北京100038

出　　处：《日用化学品科学》2022年第2期31-34,共4页Detergent & Cosmetics

基　　金：中央高校基本科研业务费项目(2021JKF204)。

摘　　要：对13种洗衣皂进行拉曼光谱的数据收集,对拉曼光谱数据预处理后,利用5种不同的机器学习方法对于不同品牌洗衣皂进行训练和测试。训练集中5种模式识别方法,准确性从高到低依次为:线性判别、子空间判别、最近邻、决策树、支持向量机,其中线性判别和子空间的识别率均高于95%,训练效果较好。测试集中线性判别和子空间判别识别率高于90%,其中线性判别识别率为100%,其余3种方法识别率均低于50%。通过综合比对,线性判别的方法适合对于不同品牌洗衣皂的识别。The Raman spectra of 13 kinds of laundry soap were collected and preprocessed.Five different machine learning methods were used to train and test for different brands of laundry soap.The accuracy of the methods in the 5 modes in the training set from high to low is:LDA,ESM,KNN,DT,SVM,among which the recognition rates of LDA and ESM are both higher than 95%,and the training effects are better.In the test set,the recognition rates of LDA and ESM are higher than 90%,in which the recognition rate of LDA is 100%,and the recognition rates of the other three methods are all lower than 50%.Through comprehensive comparison,the method of LDA is suitable for the identification of different brands of laundry soap.

关 键 词：洗衣皂 拉曼光谱 模式识别 

分 类 号：TQ649[化学工程—精细化工]