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

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.