小波变换-偏最小二乘法用于三种食用色素的可见分光光度法同时测定  

Application of wavelet transform-partial least square method for simultaneous determination of three food pigments in mixed solution using spectrophotometry.

作　　者：司圣柱[1] 司娲[2] 

机构地区：[1]安徽教育学院化学系,合肥230061 [2]复旦大学环境科学与工程系,上海200433

出　　处：《分析仪器》2007年第2期48-51,共4页Analytical Instrumentation

摘　　要：对胭脂红、苋菜红、日落黄三种食用色素混合物溶液的可见分光光度数据进行了小波变换处理,将原始吸光度数据及其小波变换系数用偏最小二乘法分析。结果表明,基于小尺度的小波低频系数的模型优于原始吸光度数据的全谱模型。用Daubechies4小波对原始吸光度数据进行一次分解,以低频系数作校正集并用交叉验证法选择主成分数进行偏最小二乘法建模,获得了令人满意的预测结果。The visible absorption spectral data of a mixed solution of three food pigments （carmine, amaranth and sunset yellow） was pretreated by wavelet transform （WT）, and the original spectral data and their wavelet transform coefficients were treated by partial least square （PLS） regression. The results showed that the model based on wavelet low frequency coefficients of small scale is better than the model based on spectal data of full spectra. The original absorption spectral data were resolved by Daubechies4 wavelet transform. The low frequency coefficient were used as calibration set, the number of principal components was selected by cross-validation method and the multivariate calibration model was established by PLS regression. The results of prediction were satisfactory.

关 键 词：小波变换 偏最小二乘法 食用色素 可见分光光度法 同时测定 

分 类 号：O657.3[理学—分析化学]