Combining blind source analysis and Elman recurrent neural network to determine overlapping voltammograms

A novel method named ICA-ERNN approach based on independent component analysis (ICA) as pre-processed tool with Elman recurrent neural network (ERNN) regression was proposed for the simultaneous differential pulse voltammetric determination of o-nitroaniline, m-nitroaniline and p-nitroaniline with overlapping peaks. The method combines the ideas of ICA with ERNN regression for enhancing the ability in the extraction of characteristic information and the quality of regression. A program (PICAERNN) was designed to perform the simultaneous voltammetric determination of o-nitroaniline, m-nitroaniline and p-nitroaniline. The relative standard errors of prediction (RSEP) obtained for all components using ICA-ERNN and ERNN were compared. Experimental results demonstrated that the ICA-ERNN method had better result than ERNN methods and was successful even when there was severe overlap of voltammgrams.