Arithmetic Of Signal Preprocessing And Fuzzy Neural Network In Recognition Of Electronic Nose

Caused by the physical shortcomings of gas sensors, it is impossible for a single gas sensor to identify multiple gases. So the electronic nose techniques based on gas sensor array and pattern recognition is becoming an important way in dealing with cross-sensitivity in gas analysis. The pattern recognition technology plays the crucial role to the behavior of electronic nose. Aimed to this requirement, the research studies several related pattern recognition techniques in order to realize the quantitatively recognition of the mixed gases.The research studies the Principal Components Analysis and Independent Component Analysis arithmetic. The principal component analysis could substitute a small number comprehensive variable for original multidimensional variable and obviously simplify data structure in the pre-condition of minimizing loss of original data information. The research combines the Principal Components Analysis and Independent Component Analysis in order to realize an effective data pre-processing in the electronic nose system. Some experiments are performed to verify the optimizing effect of PCA and ICA. Aimed to the neural network and the fuzzy logic of the pattern recognition technique, combines the logical expression ability of the fuzzy system and the self-studing ability of the neural network, brings forward a fuzzy neural network arithmetic: Takagi-Sugeno fuzzy neural network to realize the quantitatively recognition of the alcohol thickness. Based on this method, brings forward an advanced Takagi-Sugeno Fuzzy Logic System based on Neural Networks, it uses the neural network to simulate the subjective selection. This method can avoid the low efficiency and imprecise of the subjective selection and obtain the associated membership function and reasoning regulation accurately.The research uses the alcohol electronic nose testing date as the swatch, using the arithmetic mentioned above to simulate and test 54 kinds of mixed gases, the accurate rate of the quantitatively recognition can achieve to above 99%. The result proves that the signal pre-processing techniques and the fuzzy network arithmetic perform well in the alcohol' s quantitative analysis.