Abstract: Neural networks(NN) and multiwavelength spectroscopy were investigated systematically for multicomponent analysis. In this paper, neural networks and chelating spectrophotometry were applied to multivariate calibration and spectral resolution. The neural networks(NN) were trained by ordinary propagation(OBP) and the modified propagation(MBP). The chelating photometric reagent was selected as tribromochlorophosphanazo(TBCPA). The proposed sensitive spectrophotometry combined with the propagation neural networks (BPNN) was used for the simultaneous determination of 15 rare earth elements with good results. The relative standard divrations were leas than and about equal to 5%, RSD ≤ 5%, in general. It was shown that the neural networks can be used as a novel powerful chemometric technique for multicomponent analysis, especially for multivariate calibration and spectral resolution.

Key words: neural networks, backpropagation, rare earth elements, chelating spectrophotometry, Multivariate resolution and calibration, Chemometrics