Determination and Identification of Sudan IV Using Fluorescence Spectrometry and Artificial Neural Networks

Abstract

The fluorescence spectra of the solutions of industrial color Sudan IV are measured experimentally, excited by 345nm light. A wavelet transformation-radial basis function neural network is established, using the spectral data, to determine the concentrations of the Sudan IV solutions. The relative error turns out to be 0.73%. Meanwhile, the spectra of six synthetic food colors Ponceau 4R, Amaranth, Allura Red, Acid Red, Erythrosine and New Red are measured, and the wavelength of excitation light is 310nm. Another wavelet transformation-radial basis function neural network is established to identify the seven red colors mentioned above. It is shown that this method, which combines the advantages of both fluorescence spectrometry and artificial neural network, can realize accurate determination of Sudan IV and identification of industrial colors and synthetic food colors.