Odor classification of wines by using neural networks

This paper considers an application of odor classification of wines by using an electronic nose system based on a neural network. The neural network used here is a competitive neural network by the learning vector quantization. Two kinds of wines are classified by using odor data measured with an array of many metal oxide gas sensors. After bubbling two kinds of wines, the odor gases are absorbed to a spongy material called as a mono trap for a while. Then dry air is passed through the mono trap and odors are transmitted to an array of odor sensors. The gas densities are transformed into voltages and the data are stored in a computer. Then using an electronic nose system we classify the kind of wines. The accuracy rates for white wine and red wine are around 97.4% and around 83.4%, respectively.