小波神经网络结合紫外光谱用于四种维生素混合体系的同时测定

Simultaneous Determination of Four\|component B\|group Vitamins by Using Wavelet Neural Network (WNN)

首次实现了小波神经网络多节点输出， 表明该网络在处理多组分混合物的定量分析问题上有独到的优点。运用小波神经网络对 Ｂ族维生素的混合体系进行了测定， 结果良好。

A simultaneous determination of four components of B\|group vitamin, using a novel wavelet\|based neural network, wavelet neural network (WNN), combined with the correlation coefficient and standard deviation, is reported in this work. 11 representative wavelength points are selected from each original UV spectral data, based on correlation coefficients and standard deviations of the observed data. A family of subsets with a finite compact frame is built to solve the problems of high redundancy and training difficultly involved in an adaptive wavelet neural network (AWNN). The predicted results, with the fitting relation coefficient (R=0 99737) and the standard deviation (SD=0 30254), are satisfactory.