The Study Of Electronic Tongue Used In The Recognition Of Beverage

The electronic tongue is an intelligent detection system which can analyze the characteristic information of different liquids and identfy them. It is composed of three parts which are the sensor array, the data acquisition module and the pattern recognition module. The sensor array is used to obtain the characteristic information of the liquid and then the acquired information is collected by the data acquisition module and stored in the computer. Finally the data are processed by the pattern recognition algorithms. The electronic tongue has been widely used in food, medicine, environment and other fields. It has a broad application prospect.A voltammetric electronic tongue system has been developed in this paper. The front sensor array is composed of the standard three-electrode system. The differential pulse voltammetry is tried to be used as an analytical method in the system. A method of characteristic extraction is proposed to extract the useful information from the voltammetry curves obtained from the electronic tongue. The results show that the method can not only reduce the burden of data processing but also maintain the recognition accuracyThe voltammetric electronic tongue system is applied to the experiment of distinguishing five kind of green teas which are from different places, then the principal component analysis, the probabilistic neural network and the C-means clustering algorithm are used to process the experimental data. The results show that all of the results are satisfactory and because the principal component analysis and the C-means clustering algorithm don't need to train the network, the speed of recognition by them is better than the speed of recognition by the probabilistic neural networkThe potentiometric electronic tongue is used in the experiment of distinguishing five different brands of orange juice, and the experimental data were processed by the above pattern recognition algorithms. The results show that the principal component analysis combining the probabilistic neural network can classify them well, which improves the recognition accuracy for this kind of subjects and the performance of potentiometric electronic tongue system.The above three algorithms of pattern recognition are also applied to process the data of five different brands of mineral water from the potentiometric electronic tongue. The results show that the score plot drawn by three principal components has a better recognition result than the score plot drawn by two principal components in the data processing by the principal component analysis. The recognition accuracy of the probabilistic neural network can reach 100% when the number of training samples is 200, and the c-means clustering algorithm can distinguish the five brands of mineral water well.