Research On The Electrical Nose System For Liquors Recognition

Nowadays, the quality of the drink food is mainly judged by the sense of the professional people as a result, the results differ from each other with the judger's physical status, mood and environment. The electrical nose is a kind of new detective and comprehensive technology, It adopting the gas sensor array combines the pattern recognition algorithms, it can analyze the composition of the mixed gases qualitatively or quantitatively, so it has great application prospect in the fields of food manufacturing and detection, medical diagnosis, toxic gas detection and controlling. Thus we use the electrical nose to analysis various domestic liquor qualitatively.This paper, from the 65 gas sensors detect the six-month-old screening of five faster response and recovery time, sensitivity is good, sensitive to different characteristics, long-term stability, and integration of temperature and humidity sensor with integral sensor array. Electronic nose system's hardware circuit mainly completed the extraction and processing of the gas sensor array signal, using ARM Cortex-M3 processor, embedded control systems, the use of internal resources and a variety of ARM the external interface technology to achieve control of peripheral circuits, signal acquisition, data processing, algorithm implementation, and identification display. The use of Visual C + + 6.0 has developed a Windows-based user interface, by communicating between the serial-ports and pc. Shows the results of real-time signal acquisition and processing for neural networks, the online study and the parameters of the algorithm are finished, and the algorithms are downloaded on the chips to recognize gas.Because of the disadvantages in the BP algorithm, various improved BP algorithms used in the liquor qualitative analysis are analyzed and compared, among which the multiple BP subnets have a better result. Besides, the temperature and humidity, main factors influencing the performance drift, are compensated and acted as signatures with correspondence signals of other sensors. They are calculated in the Artificial Neural Network and restrained the drift to some degree; the performance of the network is also improved.The developed electrical nose has many advantages, such as small volume, low weight, good stability and low power. It realized correct reorganization on five domestic liquors, recognition rate of 96.5%.