Application of electronic tongue to discriminate white wines originated from different regions of Hungary

In this work, the application of an electronic tongue (ET) based on a specific ion-selective sensor array for discrimination of different white wine types is presented. The electronic tongue equipped with specific sensor array containing seven IFSET sensors was used to analyze wine samples. The obtained ET responses were evaluated using different pattern recognition methods. Principal component analysis (PCA) provides the possibility to identify some initial patterns. Linear discriminant analysis (LDA) was used to build models to separate white wine samples based on wine regions and grape cultivars. The results showed that every group was distinguished from each other with no misclassification error. Furthermore, the sequence of the wine sample groups was similar to the increasing total acidity content. Partial least square (PLS) regression was used to build models for the prediction of the main chemical compositions of the wine samples based on the electronic tongue results. The closest correlation (R2=0.93) was found in case of ‘total acidity’, and the prediction error (RMSEP) was 6.9%. The pH of the wine samples was predicted with good correlation (R2=0.89) but higher prediction error (RMSEP=10.71%) from the electronic tongue results. The ET combining these statistical methods can be applied to determine the origin and variety of the wine samples in easy and quick way