Simultaneous Determination of Multicomponent Nitrobenzene Compounds by Artificial Neural Networks and Visible Spectroscopy

Simultaneous determination of nitrobenzene and p-chloronitrobenzene were studied by artificial neural networks and visible spectroscopy in the paper. The effects of 2,4- dinitrochloro-benzene, p-nitrotoluene, m-nitrotoluene and the industrial wastewater matrix from the production of dye and pharmacy on the determination were mainly studied. The research indicated that when the common training method of BP algorithm was applied to the treatment of composed samples, the result was satisfactory. However, when it was applied to the determination of wastewater matrix, the error was large. But the biased training method proposed in this paper could improve the determination accuracy obviously.