摘要
为准确、快速地测定南方猪舍的主要有害气体NH3、H2S,建立了电子鼻系统。在实验室中采用静态配气法配制各种浓度的气体,将快速独立成分分析与径向基神经网络两种方法相结合,对6.95~69.53mg/m3浓度范围内的H2S单一气体以及H2S与NH3组成的混合气体进行定量识别,平均识别精度分别达到99.1%和90.97%。结果表明在基于电子鼻的猪舍NH3、H2S气体定量识别中,采用该种方法具有良好的效果。
In order to establish an accurate and rapid method for determination of the main pernicious gases NH3 and H2S in the piggery in South China,an electronic nose system was built.Gas samples were prepared with static volumetric method,and the fast independent component analysis(FICA)and the radial basis function neural networks(RBFNN)were applied to identify H2S gas alone and the mixed gas of H2S and NH3 between the range of 6.95-69.53 mg/m^3.The average identification accuracy of single gas reached 99.1%,while the average identification accuracy of mixed gas was 90.97%.The results show that good effect was obtained by applying FICA and RBFNN methods to quantitative identification of gases NH3 and H2S in piggery based on electronic nose system.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2009年第7期153-157,共5页
Transactions of the Chinese Society of Agricultural Engineering
基金
广东省科技计划攻关项目(2007A020300010-7)
华南农业大学校长基金项目(2007X024)
关键词
气体识别
径向基神经网络
传感器
电子鼻
猪舍
快速独立成分分析
gas identification
radial basis function neural networks
sensors
electronic nose
piggery
fast independent component analysis