The Study And Application Of Neural Network In Artificial Olfactory Information Processing Techniques

The artificial olfactory system based on gas sensitive sensors array and patterns recognition technology is an important method to recognize gases use the cross sensitivity of gas sensitive sensor. Main study of this paper is neural network application in recognition of artificial olfactory system. Among these, BP neural network, SOM neural network and Neuro-Fuzzy TSK network were introduced and studyed in detail.Firstly, the simple history and the research status is summarized in this paper. Then different classification methods of patterns recognition of artificial olfactory system are introduced and their advantages and disadvantages are compared.Later, an artificial olfactory system made by ourselves is introduces and is used to recognize different gases with single neural network(BP and SOM) patterns recognition method. Auto-adjusting of study parameters is used in algorithm in order to accelerate its studying speed.Finally, the combination of fuzzy inference technology and neural network is introduced. Its presentation include the structure and arithmetic of Neuro-Fuzzy TSK network and FCM which used in procedure of making inference rules. Moreover, there are tow TSK model used in recognizing different gases, and analysis of its experimental error is discussed.