Application of the wavelet transform coupled with artificial neural networks for quantification purposes in a voltammetric electronic tongue

Abstract

This work describes a voltammetric electronic tongue, in which the quantitative information contained in voltammograms obtained from amperometric sensors is firstly extracted employing the discrete wavelet transform (DWT) and then processed employing artificial neural networks (ANNs). The analytical case studied is the direct determination of the oxidizable aminoacids tryptophan, cysteine and tyrosine, and its application in the direct measurement of these amino acids in animal feed samples. A conventional voltammetry cell with a Pt working electrode is the experimental set-up and differential pulse voltammetry the selected technique. Due to the complexity of the obtained signals, the DWT pre-treatment was needed in order to eliminate noise components and compress voltammograms by selecting and extracting significant information. The ANN was subsequently used to model the system departing from the reduced information, and obtaining the concentrations of the considered species. Best results were obtained when using two hidden layers in a backpropagation neural network trained with the Bayesian regularization algorithm.