Drift correction of electronic tongue responses

In this article. drift correction algorithms were used in order to remove l inear drift in multivariate spaces of two data sets obtained by an electron ic tongue based on voltammetry. The electronic tongue consisted of various metal electrodes (Au, Ir, Pt, kh) combined with pattern recognition tools, such as principal component analysis. The first data set contained differen t types of liquid, from well defined to more complex solutions. The second data set contained different black and green teas. Component correction (CC ) was compared to a simple additive correction. In CC. the drift direction of measured reference solutions in a multivariate space was subtracted from other types of solution. In additive correction. responses from reference samples were subtracted from other samples. CC showed similar or better per formance in reducing drift compared to additive correction for the two data sets. The additive correction method was dependent on the fact that the di fferences in between samples of a reference solution were similar to the ch anges in between samples of other liquids, which was not the case with CC.