خسرو محمدی

New complexes of Cu2 , Hg2 , and Cd2 with a recently synthesized Schiff base derived from 2-(3-hydroxy-1-methylbut-2-enylideneamino)pyridine- 3-ol were applied for their simultaneous determination with artificial neural networks. A new analyticalmethod using principal component-feed forward neural networks (PC-FFNNs) and principal component-radial basis function networks (PC-RBFNs) was used. Spectral data was reduced using principal component analysis and subjected to ANNs. The data obtained from synthetic mixtures of metal ions were processed by PCFFNNs and PC-RBFNs. Performances of the proposed methods were tested with regard to relative standard error of prediction. Limit of detections and limit of quantifications were determined. The results obtained by PC-FFNNs and PC-RBFNs were compared to each other. Under the working conditions, the proposed methods were successfully applied to simultaneous determination of Hg2 , Cu2 , and Cd2 in different water and soil samples. Concentrations of metal ions in the samples were also determined by flame atomic absorption spectrometry (FAAS) and standard addition method. The amounts of metal ions obtained by the proposed methods were in good agreement with those obtained by FAAS and standard addition method.