人工嗅觉系统设计初步_王昌龙.pdfVIP

·62 · 　　　　　　　　　　　　　　　　2003 5 , (东南大学仪器科学与工程系, 江苏南京210018) 　　[　] 　采用纳米薄膜气体传感器构成传感器阵列, 采用分压式取样电 测量传感器在气 氛中电阻的响应曲线。提出“气味谱”的概念,用误差反传神经网络和自组织特征映射神经网络 (SOM)进行学习训练和识别, 结果表明SOM 网络更接近于人类对气味的认知过程, 是人工嗅觉系 统较为理想的算法。玻璃基片等离子溅射镀膜方法是制造人工嗅觉所需的大规模气体传感器阵列 的有效方法。 []　人工嗅觉;气体传感器;SOM + []TP242.6 4 　[]A 　[]1000-0682(2003)05-0062-03 The design of an artificial olfactory sytem WANG Chang-long, HUANG Wei-i (I nstrumental Science Engineering D ep t of Southeast University ,J iangsu-Nanj ing 210018, China) 　　Abstract:The arra of gas sensors is composed of nanometer film gas sensors.The resistance response curves in the atmosphere are measured b voltage dividing circuits.The paper puts forward the conception of odor spectrum along with the methods of learning,training and identif ing b means of the BP and SOM neural networks.The application shows that the SOM is nearer to the recognizing process of human being to the odor. It is the ideal algorithm of a man-made olfactor s stem.A large-scale gas sensor arra for use in the olfactor s stem can be made easil with the ion plating film on a glass substrate. Key words:Artificial olfaction;Gas sensor;SOM [3] ,, 1 　　 。:① 　　, , 、,。② , 。 , ,、() 。③、 。 1964 Wilkens Hatman , 、 ,——— 。[4] , 、 2 　 [1,2] 。 [5] , 　　 (2—1): 。 ———, , ,、,、 ,, 。 , (), 　　:2002-12-02