摘要
利用半导体气体传感器的交叉敏特性 ,将气体传感器阵列与神经网络相结合 ,构建了一个用于临场感机器人的人工嗅觉系统 ,用于气体的定性识别 .自组织神经网络 (SOM )将被测气体的多维特征信息映射到一个二维平面上 ,从而实现了对被测气体的识别分类 .实验结果表明半导体阵列人工嗅觉系统可以提高气体传感器的选择性 。
By using the overlapping sensitivity of chemical gas sensors, an artificial olfactory system based on a telepresence robotic system, combing a chemical gas sensor array with self organizing map (SOM) neural networks, is constructed to qualitatively analyze chemical gases. SOM neural networks have the remarkable capability of transforming the hyperspace characteristics of input gas into a two dimensional map, consequently the gas can be discriminated. Experimental results show that the system increases the selectivity of gas sensors and the SOM neural network is feasible for gas discrimination.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2004年第1期28-31,共4页
Journal of Southeast University：Natural Science Edition
关键词
气体传感器阵列
自组织特征映射
神经网络
人工嗅觉
gas sensor array
self organizing map
neural networks
artificial olfactory system
