基于中红外光谱技术检测茶叶中非法添加滑石粉的研究

发布时间：2019-05-29 18:52

【摘要】：研究傅里叶变化红外透射光谱(FTIR)结合化学计量学方法检测茶叶中非法添加滑石粉的可行性。首先获取掺杂量不同的滑石粉(0.00,0.15,0.25,0.35,0.50,0.65,0.75,0.85,1.00,1.10,1.25,1.50mg·g~(-1))210个茶叶样本光谱。为了突显出光谱中的细微变化,采用Savitzky-Golay(SG)平滑、标准化和标准正态变量(SNV)三种方法对原始光谱进行预处理。其中,处理效果最好的是SNV方法。随后探究光谱与掺杂量之间的定量关系,采用反向间隔偏最小二乘法(biPLS)和连续投影算法(SPA)的结合进行特征波数的选择,最终选出来5个特征波数。并利用偏最小二乘回归算法(PLS)和最小二乘支持向量机算法(LS-SVM)建立基于这5个特征波数的回归模型。其中LS-SVM模型具有更高的相关系数(RP=0.921)和更小的均方根误差(RMSEP=0.131),所以该模型具有更好的稳定性及更高的预测能力。综上所述,红外光谱技术可以定量地检测出茶叶中非法添加的滑石粉。
[Abstract]:The feasibility of detecting talc powder illegally in tea by Fourier transform infrared transmission spectroscopy (FTIR) combined with chemometrics was studied. The spectra of 210 tea samples with different doping amounts of talc powder (0.00, 0.15, 0.25, 0.35, 0.50, 0.65, 1.05, 1.00, 1.10, 1.25, 1.50mg 路g ~ (- 1) were obtained. In order to highlight the subtle changes in the spectrum, Savitzky-Golay (SG) smoothing, standardization and standard normal variable (SNV) are used to preprocess the original spectrum. Among them, the best treatment effect is the SNV method. Then the quantitative relationship between spectrum and doping quantity is explored. The combination of reverse interval partial least square method (biPLS) and continuous projection algorithm (SPA) is used to select the characteristic wavenumber, and finally five characteristic wavenumber are selected. The partial least square regression algorithm (PLS) and the least squares support vector machine (LS-SVM) algorithm are used to establish the regression model based on these five characteristic wavenumber. The LS-SVM model has higher correlation coefficient (RP=0.921) and smaller root mean square error (RMSEP=0.131), so the model has better stability and higher prediction ability. To sum up, infrared spectroscopy can quantitatively detect talc powder added illegally in tea.
【作者单位】：浙江大学生物系统工程与食品科学学院;
【基金】：国家自然科学基金项目(61201073);国家自然科学基金项目(31471417) 浙江省科技计划项目(2015C02008) 浙江省教育厅科研项目(Y201225966)资助
【分类号】：TS272.7;O657.33

【相似文献】