Design Of Electronic Nose/Tongue System And Application Of Data Fusion Technology

Electronic nose is a kind of instrument to imitate the mamalian olfactory, whichis composed of array of chemical sensors with partial selective and appropriatepattern recognition algorithm. It can identify simple or complex odors. Electronictongue is a kind of instrument to imitate the mamalian taste. It can evaluate theoverall taste of the liqued sample rapidly. The separate use of the technology of theelectronic nose or electronic tongue to detect the substances can only get the singleevalution of odor or taste. It cannot get the whole information to describe the object.At the same time the sensors of the electronic nose and the electronic tongue arelimited, it can not be achieved for different test objects using different sensor arraysfor detection. So it is very necessary to design different fusion system of electronicnose and electronic tongue to detect different substances.This paper designed two kinds of electronic nose systems based on differentsensor arrays. They are respectively the electronice nose system based on the TGStype gas sensors and the electronic nose system based on the MQ、MP type gassensors. The former use8TGS seris gas sensors as the sensor array and the latter use8MQ、MP seris gas sensors as the sensor array. At the same time, the design of theelectronic nose soft system is based on Labview. In addition, we built voltammetricelectronic tongue based on a standard three electrode system. The working electrodeis a gold electrode; the reference electrode is a saturated calomel electrode; and theauxiliary electrode is a platinum electrode. Through the feature-level fusion methodsof the integration information, namely fusion the system of electronic nose based ondifferent sensor arrays and the electronic tongue, and then we get two different fusionsystem of the electronic nose and the electronic tongue.These two different fusion systems of the electronic nose and electronic tongueare used to identify eight different brands of liquor and six different flavors of tea. We use principal component analysis, k-means and support vector machine to anaylze thetest data, and the two recognition results are contrasted. The results show that:compares to only using electronic nose or electronci tongue, when we detect theliquors, the recognition result of the fusion system of the electronic nose based onMQ、MP type gas sensors and the electronic tongue is the best. And when we detectthe milk tea, the recognition result of the fusion system of the electronic nose basedon TGS type gas sensors and the electronic tongue is the best.