微含量简单成分气体的人工嗅觉分析方法

提出了由两层径基函数(RBF)网络和两层线性基本函数(LBF)网络组成的串联神经网络,建立了由SnO₂半导体气敏传感器阵列和串联神经网络分类方法组成的人工嗅觉系统.对微含量的甲醇、乙醇、丙酮挥发气的测试与分析结果表明,提高了人工嗅觉系统的灵敏度和识别能力,为人工嗅觉系统对简单与复杂成分气味的实时检测和食品香气质量的人工分析创造了条件.关键词嗅觉模拟,串联神经网络,气敏传感器阵列.

Artificial Olfactory Analyses of Micro- and Simple- Component Vapors

A cascade neural network with radial basis function(RBF) and linear basis function(LBF) is proposed, and an artificial olfactory system made up of fourteen of SnO 2 model semiconductor sensors and a cascade neural network classification method set up. Measurements and analyses for eight kinds of micro and simple methanol vapors as well as ethanol and acetone ones show that with the help of the presented method an artificial olfactory system has its performances such as sensitivity and discriminating ability greatly developed, that it is able to real time detect simple component odors and multi component ones. This development lays down a foundation for such a system to determine qualities of food fragrances.