【摘要】：利用共聚焦拉曼光譜技術(shù)對(duì)茶葉中非法添加的重金屬染料——美術(shù)綠進(jìn)行檢測(cè)研究。首先通過(guò)特定的濃縮方法,獲取了五個(gè)濃度水平美術(shù)綠茶湯樣本的拉曼光譜。通過(guò)比對(duì)標(biāo)準(zhǔn)品拉曼光譜,對(duì)混有美術(shù)綠的樣本光譜進(jìn)行了定性分析。并找到了能夠用于定性鑒別茶葉中美術(shù)綠的4個(gè)主要拉曼特征波數(shù),分別為1 341,1 451,1 527和1 593cm~(-1)。對(duì)原始拉曼光譜進(jìn)行預(yù)處理后,融合反向間隔偏最小二乘(biPLS)、競(jìng)爭(zhēng)性自適應(yīng)重加權(quán)算法(CARS)和連續(xù)投影算法(SPA)對(duì)拉曼光譜中美術(shù)綠的特征波段進(jìn)行深入挖掘,最終優(yōu)選出了14個(gè)特征波數(shù)�；谶@14個(gè)特征波數(shù)分別建立了偏最小二乘(PLS)回歸模型和最小二乘支持向量機(jī)(LS-SVM)模型,結(jié)果表明,兩類模型均具有好的穩(wěn)健性和很高的預(yù)測(cè)能力,模型的建模集、驗(yàn)證集和預(yù)測(cè)集的決定系數(shù)(R~2)均超過(guò)了0.9,證明了所提取出來(lái)的特征波數(shù)的有效性。與偏最小二乘回歸模型相比,基于LS-SVM的非線性定量檢測(cè)模型的效果更佳,預(yù)測(cè)集決定系數(shù)(R~2)達(dá)到0.964,均方根誤差(RMSE)為0.535。以上研究結(jié)果表明,共聚焦拉曼技術(shù)結(jié)合特定的樣品處理方法及化學(xué)計(jì)量學(xué)方法,可以實(shí)現(xiàn)茶葉中非法添加美術(shù)綠的定量檢測(cè)。該研究為茶葉中非法添加美術(shù)綠這一食品安全問(wèn)題的有效監(jiān)管提供了幫助。
[Abstract]:The confocal Raman spectroscopy was used to detect the heavy metal dyestuff in tea. Firstly, the Raman spectra of five artistic green tea soup samples were obtained by a specific concentration method. By comparing the Raman spectra of standard samples, the spectrum of samples mixed with fine arts green was qualitatively analyzed. The four main Raman characteristic wave numbers which can be used for qualitative identification of fine arts green in tea are 1 341 ~ (-1) C ~ (-1) and 1 593 cm ~ (-1), respectively. After pretreatment of the original Raman spectrum, the feature bands of the fine arts green in the Raman spectrum are deeply mined by combining the reverse interval partial least square (biPLS), competitive adaptive reweighting algorithm (CARS) and the continuous projection algorithm (SPA). Finally, 14 characteristic wavenumber were selected. Based on the 14 characteristic wavenumber, the partial least squares (PLS) regression model and the least squares support vector machine (LS-SVM) model are established, respectively. The results show that both models have good robustness and high predictive ability. The determinant coefficients (R _ (2) of both the verification set and the prediction set are higher than 0.9, which proves the validity of the extracted characteristic wavenumber. Compared with the partial least square regression model, the nonlinear quantitative detection model based on LS-SVM is more effective. The prediction set determination coefficient (RG-2) is 0.964, and the root mean square error (RMSE) is 0.535. The results show that confocal Raman technique combined with specific sample treatment and chemometrics can be used to detect the illegal addition of fine arts green in tea leaves. The study helps to regulate the food safety problem of illegally adding art green to tea.
【作者單位】： 浙江大學(xué)生物系統(tǒng)工程與食品科學(xué)學(xué)院;
【基金】：國(guó)家自然科學(xué)基金項(xiàng)目(61201073,31471417) 浙江省教育廳科研項(xiàng)目(Y201225966) 浙江大學(xué)基本科研業(yè)務(wù)費(fèi)專項(xiàng)資金項(xiàng)目(2015QNA6005)資助
【分類號(hào)】：O657.37;TS272.7

【相似文獻(xiàn)】

相關(guān)期刊論文 前2條

1 ;017美術(shù)綠質(zhì)量改進(jìn)試驗(yàn)總結(jié)(第二部分 擴(kuò)大試驗(yàn)及試生產(chǎn))[J];廣州化工;1978年02期

2 ;[J];;年期

，

本文編號(hào)：2325612