Abstract:

In order to solve the application problems in food quality evaluation by electronic nose, an embedded electronic nose based on metal oxide semiconductor gas sensors array and VI (Virtual Instrument) Labview software was designed. The identification ability of the device was investigated by employing different categories of simple and complex odor databases. A data acquisition system based on ARM-STM32 microprocessor and 24-bit ADC (AD7794) was also presented. The test data were uploaded to PC (Personal Computer) through RS232 communication mode. Then Matlab software was used to the organization of test data. Finally the fingerprint information of the target gas was obtained by the PCA (Principal Component Analysis). In our experiment, five different sauces were tested though the volatile gases, which were derived from sauces. The original database was built. Then comparing with the database, the unknown sauce could be recognized well. The results show that this system has high accuracy in the classification of gas mixtures.

Key words: E-nose, pattern recognition, Labview, principal component analysis(PCA)