Five semiconductor gas sensors sensitive to CH4

C3H8 and H2 are chosen to compose the gas sensor array

and an on-line data acquisition system is constructed.Combining with the algorithm of feature extraction and pattern recognition

an electronic nose system is developed for the real-time detection of the three combustible gases.The method of double neural network is proposed for quantitative analysis of multi-gas which is unknown.The first network is for gas qualitative analysis

and the second one is for gas quantitative analysis.Based on BP neural network analysis

the recognition rate of the three gases above is 100%

and the maximum relative error of quantitative analysis is not more than 9.4%.