摘要
采用6只不同掺杂的纳米ZnO气体传感器组成的电子鼻实现了空气中乙醇、丙酮、苯的定量分析。对比了多元线性回归(MLR)、主成分回归(PCR)、人工神经网络(ANN)等不同算法在本实验中的应用,其体积分数平均标准误差分别为16.71×10-6,34.183×10-6,7.242×10-6。基于非线性函数映射的ANN算法具有最佳定量精度。研究表明:电子鼻在可挥发有机物(VOCs)定量分析中有潜在的应用前景。
Quantification of ethanol, acetone and benzene is analysed by using an electronic nose which contained six gas sensors doped nano ZnO. Multiple linear regression （ MLR ）, principal component regression （ PCR ） and artificial neural network （ANN） are compared for their suitability on quantifying VOCs. The average standard deviation of the volume fractions of VOCs is 16.71 ×10^-6,34. 183 ×10^-6 and 7. 242 ×10^-6 ,respectively. ANN is the most suitable algorithm in the experiment because it can realize the nonlinear mapping. This study shows the potential application of the electronic noses for quantifying VOCs.
出处
《传感器与微系统》
CSCD
北大核心
2006年第11期21-24,共4页
Transducer and Microsystem Technologies
关键词
电子鼻
可挥发有机物
定量
electronic nose
volatile organic compounds （VOCs）
quantification
