摘要
针对差分吸收光谱技术(DOAS)获取差分吸收光学密度(OD)的技术特点,本文提出采用自适应神经模糊推理系统(ANFIS)来逼近DOAS测量谱中慢变成分,并依次获得痕量气体差分吸收光学密度。该方法根据DOAS系统获得测量谱的特点进行自适应调整网络参数,因此能较精确地分离慢变部分和快变部分。利用SO2样气进行实验,结果表明ANIFIS比多项式函数更为灵活。基于AINFIS网络获取的OD进行浓度反演,提高了DOAS反演精度。
According to the characteristic of Differential Optical Absorption Spectroscopy (DOAS) system, the Adaptive Neural Fuzzy Inference System (ANFIS) was proposed to evaluate the slowly varying components by using the nonlinear approaching theory of ANFIS in this paper. The network parameter can be flexibly regulated according to the measuring spectra. The testing results show that the ANFIS is more flexible than polynomial fitting and the fitting accuracy can be easily solved. The reliability and accuracy of DOAS evaluation have been improved for the application of the ANFIS.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2008年第12期50-53,共4页
Opto-Electronic Engineering
基金
安徽高校省级自然科学研究计划项目(KJ2008A114)
国家高技术研究发展计划863资助项目(2007AA12Z109)
淮北煤炭师范学院博士启动基金资助项目(2008)
关键词
DOAS
差分吸收光学密度
自适应神经模糊推理系统
慢变
逼近
DOAS
differential absorption optical density
adaptive neural fuzzy inference system
slowly varying component
approaching