小波变换-偏最小二乘法用于多组分混合物光度法同时测定

Application of Wavelet Transform-Partial Least Square Method to Simultaneous Determination of Multi-components Using Spectrophotometry

在波长范围380-600nm测定了藏红、荧光桃红、曙红三种色素混合物溶液的吸光度数据,用小波变换对光谱数据进行处理,然后将小波变换系数用偏最小二乘(PLS)法进行回归分析。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PLS建模,获得了令人满意的预测结果。

The absorbance data is determined for ternary mixture solution of safranine,phloxine B and eosin Y within wavelength ranging 380～600 nm,wavelet transform（WT）is then employed to process the obtained data of absorbance spectrum,and finally the partial least square（PLS） regression procedure is performed on the wavelet coefficients.Therefore,the original absorbance spectrum is decomposed with Daubechies4 wavelet transform,and cross-validated with low frequency coefficient as a calibration set to select the optimal number of components modeled on PLS,and the desirable prediction results is obtained.