Application of electronic nose technology in detection of combustible gas

Based on the study of the theory and constituent of the electronic nose systems, a set of combined gas sensor array, microcontroller and PC for detection of gas mixture is designed and constructed. Four kinds of gas (hydrogen, methane, acetylene, propane) are tested by the system. The feature parameters are picked up from each curve of the gas sensor's reaction. The experimental samples are analyzed by the method of PCA (principal component analysis) and BPNN (BP neural network). The result shows that the four kinds of gas could not be distinguished well by the method of PCA. While BPNN can solve the problem of sensor cross-sensitivity, and four kinds of gas can be identified qualitatively, the recognition probability of BPNN is more than 90%.