《the electronic nose》由会员分享，可在线阅读，更多相关《the electronic nose（8页珍藏版）》请在金锄头文库上搜索。
1、AU J.T. 9(1): 1-8 (Jul. 2005) The Electronic Nose A Big Part of Our Future David Tin Win Faculty of Science and Technology, Assumption University Bangkok, Thailand Abstract The four functions of the human nose involved in odor sensing: detection, recording, memory search and identification, are repl
2、icated by machines, called electronic noses or e-noses. Chemical sensors are used for detecting and recording; and artificial intelligence (AI) software (Artificial Neural Net ANN) is used for memory search and identification. Arrays of chemical sensors produce electrical signals that are integrated
3、 by advanced signal processing for rapid identification of samples. The usual detection system consists of multiple arrays of small odor sensors, such as the Metal Oxide Sensor (MOS) type, Japanese Taguchi tin dioxide SnO2 chemical sensors. Principles, possible applications and some commercial instr
4、uments are considered. Food, cosmetics, perfumes and aroma, chemistry, polymers, medical diagnostics, environmental and military are identified as some of the possible application areas. Keywords: AromaScan, Alpha MOS, MultiSampler, artificial intelligence, Artificial Neural Nets, chemical sensors,
5、DNA based memory chips, headspace, nano particles, odor sensors, Taguchi sensors, VaporLab. Introduction The human nose is a sense organ that senses odors. Experiencing smell sensation consists of the initial impacting of vapors on the sensor part of the nose, followed by the four functions: detecti
6、on, recording, memory search and identification. The initial impact is physical whereas the latter four functions of detecting and recording the electrical signal caused by the impact, searching the memory and matching the signal with previously stored signals in the memory, and finally identifying
7、the impact signal, are mental processes.1 The resultant is a response based on like or dislike. On a slightly more elaborate level, the smell sense may be seen as starting with gas particles (molecules or atoms) entering the nose and striking the olfactory organs the smell sense organs. This stimulu
8、s produces electrical signals that are recorded, matched with stored memory and finally identified as a definite sensation (Win 2000). The initial impact and the four functions may be replicated by machines. Such machines are electronic noses or e-noses. An example, the JPL e-nose, is shown in Fig.
9、4. 1 In Buddhist terminology, the initial physical impact is called rupa component and the four functions are called nama components. These instruments use chemical sensors for detecting and recording; and artificial intelligence (AI) software for memory search and identification. The common basic d
10、etection system is composed of multiple arrays of small odor sensors, such as the Metal Oxide Sensor (MOS) type, Taguchi tin dioxide SnO2 chemical sensors. Typical responses are shown in Figs. 1 and 2. The normalized response, which is normalized to a selected standard sample, is shown in Fig. 3 (Wi
11、n 2000). Modern sensor research and the development of artificial intelligence (AI) Artificial Neural Net (ANN) software such as the Brainmaker (BM) have opened up the possibility of e-noses that can effectively mimic the human nose. Technically the term Electronic Nose is a generic name for an anal
12、ytical instrument that profiles the sample 1 AU J.T. 9(1): 1-8 (Jul. 2005) headspace volatiles. Outputs from an array of chemical sensors are integrated by advanced signal processing for identification of samples. The output from sensor arrays could be analyzed with a neural network, thereby allowin
13、g an array to learn different smells on its own, and greatly enhance versatility. For instance an array could be taught what a standard environment is and what the acceptable limits are, and if the air goes outside the range in any direction, warning beeps are emitted. Similarly an array could be tr
14、ained to detect when cooking oil needs to be changed, a function that would be welcome by businesses like KFC. Thus the integration of multiple gas sensors and artificial intelligence has led to a new science of machine olfaction. E-noses may eventually replace panels of human smellers. Some commerc
15、ial instruments have been marketed since about eight years ago. The first generation of electronic noses incorporated a specific sensor. But the current trend is towards development of broadly tuned e-noses containing trainable arrays that respond differently to various smells. This paper considers
16、the principles of the e-nose; identifies possible applications; and lists some commercial instruments. Principles In e-noses, the four basic functions2 of the human nose for sensing smell are covered. Arrays of chemical sensors are used for detecting and recording functions; and artificial intelligence (AI) based Artificial Neural Network (ANN) software such as the Brainmaker (BM) is used for memory search and identification functions. Thus e-noses can effectively mimic the human
