基于快速不动点法及神经网络的机动车尾气NO和NO_2定量分析研究  

Quantitative Analysis of NO and NO_2 from Vehicle Exhaust Emission Based on Fast ICA and ANN

作　　者：张恺[1,2] 张玉钧[1] 何莹[1,2] 尤坤[1] 刘国华[1,2] 陈晨[1,2] 高彦伟[1,2] 贺春贵[1,2] 鲁一冰 刘文清[1] 

机构地区：[1]中国科学院安徽光学精密机械研究所中国科学院环境光学与技术重点实验室,安徽合肥230031 [2]中国科学技术大学,安徽合肥230026

出　　处：《大气与环境光学学报》2016年第6期435-441,共7页Journal of Atmospheric and Environmental Optics

基　　金：国家863计划;2014AA06A503;国家重大科学仪器设备开发专项;2012YQ22011902~~

摘　　要：机动车尾气对环境的危害日益加重,机动车尾气排放浓度的检测对大气污染治理具有重要意义.设计了基于非分散紫外的机动车尾气NO、NO_2浓度检测系统,搭建了实验装置,获得NO、NO_2混合气体的吸收光强后,利用快速不动点(Fast ICA)算法和人工神经网络模式识别算法对机动车尾气排放NO、NO_2组分进行定量分析.实验结果表明,利用所设计的算法对600 ppm以内的NO气体和200 ppm以内的NO_2气体浓度进行测量,其相对误差最大为1.54%,最小为0.25%。With the increasing number of vehicles, the harm from vehicle exhaust to the environment becomes more and more serious. So the monitoring of the concentration of vehicle exhaust emissions is very important to assess the emission levels. The NO and NO2 quantitative detection system based on nondispersion ultraviolet （NDUV） for vehicle exhaust emissions is built, and the original data of the mixed tail gas is obtained. And then, the identification and quantitative analysis of NO and NO2 gas is carried out with fast independent component analysis （Fast ICA） and artificial neural network （ANN） recognition algorithms. It can be drawn from the results that using the two algorithms, the NO concentration （under 600 ppm） and NO2 concentration （under 200 ppm） can be detected accurately and the maximum relative error is 1.54%, and the minimum is 0.25% .

关 键 词：机动车尾气 NO NO2 定量分析 快速不动点 人工神经网络 

分 类 号：X734.2[环境科学与工程—环境工程] X831