A guiding method to select and reduce the number of sensing units in electronic tongues

This work describes a guiding method to select the best sensing units to compose the array of an electronic tongue. It is based on the principal component analysis (PCA), in which use is made of the PCA plot of the transposed matrix of experimental data, e.g., for different concentration of an analyte. When points in such a plot appear very close to each other, it means that one of the corresponding sensing unit can be removed from the array without hampering the discrimination capability of the e-tongue. The method was validated with electric capacitance data obtained with aqueous solutions made with ultrapure water and KCl or HCl or glucose at different concentrations.