用离散变换主组分回归法处理重叠紫外吸收光谱

将离散小波变换、小波包变换、傅里叶变换和离散余弦变换与主组分回归方法结合构成4种离散变换主组分回归方法,编制了离散变换主组分回归方法的计算程序。将离散变换主组分回归方法用于处理水杨酸甲酯、邻苯二甲酸二丁酯和邻苯二甲酸氢钾混合物的重叠紫外吸收光谱数据。实验结果表明,离散变换主组分回归方法优于主组分回归方法,试样质量浓度的预测值与实际值的相对标准误差由3.94%降至2.93%,回收率由90.5%~114.8%改善为91.0%~105.5%。

Treating of Overlapping UV Spectrum by Discrete Transform Principal Component Regression

Four kinds of discrete transform principal component regression methods were established by combining wavelet packet transform,wavelet transform,Fourier transform and discrete cosine transform with principal component regression.Calculating programs were designed to perform the calculation by discrete transform principal component regression methods.The methods were applied to treat overlapping UV spectrum of mixtures of methyl salicylate,dibutyl phthalate and potassium acid phthalate.The experimental results indicated that discrete transform principal component regression methods were better than principal component regression method.Relative standard errors between actual and estimated data of sample mass concentrations decreased from 3.94% to 2.93% and recovery was improved from 90.5%-114.8% to 91.0%-105.5%.