三维荧光二阶校正与非线性回归同时测定多个有机化合物

Simultaneous determination of organic compounds by combining three-dimensional fluorescence spectroscopy,second-order calibration with nonlinear regression

由于内嵌的二阶优势,基于三维数据分解的二阶校正技术在例行分析中具有巨大的潜力。三维荧光光谱法是描述荧光强度同时随激发波长和发射波长变化的关系谱图,能提供更完整的荧光光谱信息。平行因子分解和交替三线性分解算法是对不同荧光物质的光谱进行分离的高效算法,可从复杂混合物的三维荧光光谱数据矩阵中将其各自的特征荧光光谱分解出来。邻苯二酚、对苯二酚、酪氨酸、吲哚是结构非常相似的有机化合物,光谱重叠严重,很难用常规方法直接定量检测。本实验将二阶校正算法、神经网络非线性回归与三维荧光光谱法结合,探究了四组分同时分离检测方法,实验结果确认了该方案的可行性,通过比较也发现两种二阶校正方法无明显的差异。该方法在环境分析中具有巨大潜力。

Due to the second‐order advantage ,second‐order calibration model based on parallel factor analysis （PARAFAC ） decomposition of three‐way data shows enormous potential in routine analysis . Three‐dimensional fluorescence spectrometry can describe the relationship between fluorescence intensity and the change in the wavelength of emission and excitation simultaneously .More information of fluores‐cence spectrum can be provided .Both parallel factor analysis （PARAFAC） and alternating trilinear decom‐position （ATLD） are classic algorithms and can separate the spectra of different substances with similar structure .The characteristic fluorescence spectrum can be obtained by decomposing data matrix of three‐dimensional fluorescence spectra from complex mixture .Catechol ,hydroquinone ,indole and tyrosine have exhibit similar structure .It is difficult to directly quantitative these components by a conventional method .＆amp;nbsp;By combining second‐order calibration ,the possibility of simultaneous determination of nonlinear regres‐sion of neural network and three‐dimensional fluorescence spectroscopy are investigated .The experimental results confirm the method is feasible and no obvious performance difference between PARAFAC and A T LD are observed .