Application of Neural Networks to Identify Wine Based on Electronic Tongue

The electronic tongue is a new measuring instrument, consist of the pattern recognition and an array of taste sensors which is used to examine the characteristics of liquid. Now it has a widely application in drink recognition. This article classifies three kinds of grape wine by the sensor arrays, first we use the principal components analytic method (PCA) to optimize the primitive sampled data, and then classify the data though the neural network method which bases on the BP algorithm, the RBF algorithm, the SOM algorithm and the LVQ algorithm. The experiment results indicated when the processed data are put into the neural network, the precision of recognition is improved, the training time is reduced, and the structure of neural networks becomes simple. Compared with all kinds of network algorithms, the LVQ networkpsilas test recognition rate has reached 100%, more suitable to recognize the grape wine.