Smartongue is a voltammetric electronic tongue, based on a non-specific sensors array and one special voltammetry, so called multi-frequency large amplitude pulse voltammetry (MLAPV), which made its responding signals have much overlapping information. Three non-linear multivariate data analysis methods, Kernel principal component analysis (Kernel PCA), Locally linear Embedding (LLE) and Sammon mapping, were used to dig the information from the collecting data of Smartongue. One linear data analysis method, normal principal component analysis (PCA) and the discrimination index (DI value) were applied as the reference method and as the quantitative indicator to evaluate the discrimination ability. The results indicated that three non-linear data processing methods exhibited much more feasible and efficient than PCA in Smartongue. Sammon mapping is the most suitable non-linear method to process data in Smartongue. It was able to extract the useful information from the raw data and to classify three bitter solutions, six artificial green tea products and five milk powder solutions by means of the storage time well. Sammon mapping will be a very promising data processing technique for voltammetry electronic tongue.