Mixed Gas Concentration Inversion Based on the Ultraviolet Absorption Spectrum by a Hierarchical Convolutional Neural Network

Mixed Gas Concentration Inversion Based on the Ultraviolet Absorption Spectrum by a Hierarchical Convolutional Neural Network
复制标题

基于紫外线吸收光谱的分层卷积神经网络混合气体浓度反演

DOI:
10.1007/s10812-022-01421-6
复制
发表时间:
2022-09
影响因子:
0.7
通讯作者:
Y. Cui
Y. Cui
中科院分区:
化学4区
文献类型:
--
作者:
C. Lu;Y. Bian;X. Hu;S. Jin;Y. Huang;Y. Cui

文献摘要

参考文献

相似文献

提出了一种用于混合气体浓度反演的分层卷积神经网络模型。在我们的实验中,SO2,NO2和NH3的混合物进行了分析。SO2和NO2为检测气体,NH3为干扰气体。对于模拟样品,SO2和NO2的平均绝对误差分别为0.5和0.9 ppm。对于实验样品,当组分的吸收强度相差不超过一个数量级时,模型表现良好。与没有分层结构的单模块CNN模型相比,结果表明分层结构减少了交叉干扰,在很大程度上提高了预测精度。我们相信,我们的模型将有一个很好的应用领域的气体检测。
A hierarchical convolutional neural network (CNN) model for mixed gas concentration inversion is proposed. In our experiment, mixtures of SO2, NO2, and NH3were analyzed. SO2and NO2were the detected gases, while NH3was the interfering gas. For the simulation samples, the average absolute errors were 0.5 and 0.9 ppm for SO2and NO2, respectively. For the experimental samples, the model performed well when the absorption intensities of components differed by no more than one order of magnitude. Compared with the single-module CNN model without a hierarchical structure, the results demonstrate that the hierarchical structure reduces cross-interference and improves the prediction accuracy to a great extent. We believe that our model will have a promising application in the field of gas detection.
DOI: 10.1016/j.engappai.2020.103643
发表时间: 2020-06
影响因子: 8
作者:
Zhang Yiwen;Guo Haishuai;Lu Zhihui;Lu Zhan;Patrick C.K. Hung
通讯作者: Patrick C.K. Hung
DOI: 10.1016/j.chemosphere.2020.127237
发表时间: 2020-06
期刊: Chemosphere
影响因子: 8.8
作者:
Jun-Ho Yang;Jaehun Jung;J. Ryu;J. Yoh
通讯作者: Jun-Ho Yang;Jaehun Jung;J. Ryu;J. Yoh
DOI: 10.1086/505615
发表时间: 2006-08-20
影响因子: 4.9
作者:
Cheng, Bing-Ming;Lu, Hsiao-Chi;Yung, Yuk L.
通讯作者: Yung, Yuk L.
DOI: 10.1007/s10812-020-00962-y
发表时间: 2020-03-26
影响因子: 0.7
作者:
Wang, X.;Li, Z.;Wang, W.
通讯作者: Wang, W.
DOI: 10.1016/j.atmosenv.2005.02.023
发表时间: 2005-06
影响因子: 5
作者:
Y. Matsumi;H. Shigemori;Kenshi Takahashi
通讯作者: Y. Matsumi;H. Shigemori;Kenshi Takahashi