%PDF-1.5
%
1 0 obj
<<
/Metadata 2 0 R
/Names 3 0 R
/OpenAction 4 0 R
/Outlines 5 0 R
/PageMode /UseNone
/Pages 6 0 R
/Type /Catalog
/ViewerPreferences <<
/FitWindow true
>>
>>
endobj
7 0 obj
<<
/Author (Rongxiao Wang, Bin Chen, Sihang Qiu, Zhengqiu Zhu, Yiduo Wang, Yiping Wang and Xiaogang Qiu)
/CreationDate (D:20180710092116+08'00')
/Creator (LaTeX with hyperref package)
/Keywords (hazardous gas dispersion prediction; back propagation network; support vector regression; input selection; field case)
/ModDate (D:20180918095153+02'00')
/PTEX.Fullbanner (This is pdfTeX, Version 3.14159265-2.6-1.40.18 \(TeX Live 2017/W32TeX\) kpathsea version 6.2.3)
/Producer (pdfTeX-1.40.18)
/Subject (Dispersion prediction plays a significant role in the management and emergency response to hazardous gas emissions and accidental leaks. Compared with conventional atmospheric dispersion models, machine leaning \(ML\) models have both high accuracy and efficiency in terms of prediction, especially in field cases. However, selection of model type and the inputs of the ML model are still essential problems. To address this issue, two ML models \(i.e., the back propagation \(BP\) network and support vector regression \(SVR\) with different input selections \(i.e., original monitoring parameters and integrated Gaussian parameters\) are proposed in this paper. To compare the performances of presented ML models in field cases, these models are evaluated using the Prairie Grass and Indianapolis field data sets. The influence of the training set scale on the performances of ML models is analyzed as well. Results demonstrate that the integrated Gaussian parameters indeed improve the prediction accuracy in the Prairie Grass case. However, they do not make much difference in the Indianapolis case due to their inadaptability to the complex terrain conditions. In addition, it can be summarized that the SVR shows better generalization ability with relatively small training sets, but tends to under-fit the training data. In contrast, the BP network has a stronger fitting ability, but sometimes suffers from an over-fitting problem. As a result, the model and input selection presented in this paper will be of great help to environmental and public health protection in real applications.)
/Title (Comparison of Machine Learning Models for Hazardous Gas Dispersion Prediction in Field Cases)
/Trapped /False
>>
endobj
2 0 obj
<<
/Length 5751
/Subtype /XML
/Type /Metadata
>>
stream
application/pdf
Rongxiao Wang, Bin Chen, Sihang Qiu, Zhengqiu Zhu, Yiduo Wang, Yiping Wang and Xiaogang Qiu
Dispersion prediction plays a significant role in the management and emergency response to hazardous gas emissions and accidental leaks. Compared with conventional atmospheric dispersion models, machine leaning (ML) models have both high accuracy and efficiency in terms of prediction, especially in field cases. However, selection of model type and the inputs of the ML model are still essential problems. To address this issue, two ML models (i.e., the back propagation (BP) network and support vector regression (SVR) with different input selections (i.e., original monitoring parameters and integrated Gaussian parameters) are proposed in this paper. To compare the performances of presented ML models in field cases, these models are evaluated using the Prairie Grass and Indianapolis field data sets. The influence of the training set scale on the performances of ML models is analyzed as well. Results demonstrate that the integrated Gaussian parameters indeed improve the prediction accuracy in the Prairie Grass case. However, they do not make much difference in the Indianapolis case due to their inadaptability to the complex terrain conditions. In addition, it can be summarized that the SVR shows better generalization ability with relatively small training sets, but tends to under-fit the training data. In contrast, the BP network has a stronger fitting ability, but sometimes suffers from an over-fitting problem. As a result, the model and input selection presented in this paper will be of great help to environmental and public health protection in real applications.
Comparison of Machine Learning Models for Hazardous Gas Dispersion Prediction in Field Cases
2018-07-10T09:21:16+08:00
LaTeX with hyperref package
2018-09-18T09:51:53+02:00
2018-09-18T09:51:53+02:00
hazardous gas dispersion prediction; back propagation network; support vector regression; input selection; field case
pdfTeX-1.40.18
False
This is pdfTeX, Version 3.14159265-2.6-1.40.18 (TeX Live 2017/W32TeX) kpathsea version 6.2.3
uuid:8f8e784b-9072-43ef-ba6c-f2e6b99965fe
uuid:252fd3ef-7d45-4273-9be7-c9ba9845ff74
endstream
endobj
3 0 obj
<<
/Dests 8 0 R
>>
endobj
4 0 obj
<<
/D [9 0 R /FitH]
/S /GoTo
>>
endobj
5 0 obj
<<
/Count 20
/First 10 0 R
/Last 11 0 R
/Type /Outlines
>>
endobj
6 0 obj
<<
/Count 20
/Kids [12 0 R 13 0 R 14 0 R 15 0 R]
/Type /Pages
>>
endobj
8 0 obj
<<
/Kids [16 0 R 17 0 R 18 0 R]
/Limits [(AMS.1) (table.3)]
>>
endobj
9 0 obj
<<
/Annots [19 0 R 20 0 R 21 0 R 22 0 R 23 0 R 24 0 R 25 0 R 26 0 R]
/Contents [27 0 R 28 0 R 29 0 R 30 0 R 31 0 R 32 0 R 33 0 R 34 0 R]
/CropBox [0 0 595.276 841.89]
/MediaBox [0 0 595.276 841.89]
/Parent 12 0 R
/Resources 35 0 R
/Rotate 0
/Type /Page
>>
endobj
10 0 obj
<<
/A 36 0 R
/Next 37 0 R
/Parent 5 0 R
/Title
>>
endobj
11 0 obj
<<
/A 38 0 R
/Parent 5 0 R
/Prev 39 0 R
/Title
>>
endobj
12 0 obj
<<
/Count 7
/Kids [40 0 R 9 0 R 41 0 R 42 0 R 43 0 R 44 0 R 45 0 R]
/Parent 6 0 R
/Type /Pages
>>
endobj
13 0 obj
<<
/Count 6
/Kids [46 0 R 47 0 R 48 0 R 49 0 R 50 0 R 51 0 R]
/Parent 6 0 R
/Type /Pages
>>
endobj
14 0 obj
<<
/Count 6
/Kids [52 0 R 53 0 R 54 0 R 55 0 R 56 0 R 57 0 R]
/Parent 6 0 R
/Type /Pages
>>
endobj
15 0 obj
<<
/Count 1
/Kids [58 0 R]
/Parent 6 0 R
/Type /Pages
>>
endobj
16 0 obj
<<
/Kids [59 0 R 60 0 R 61 0 R 62 0 R 63 0 R 64 0 R]
/Limits [(AMS.1) (cite.B6-ijerph-320180)]
>>
endobj
17 0 obj
<<
/Kids [65 0 R 66 0 R 67 0 R 68 0 R 69 0 R 70 0 R]
/Limits [(cite.B7-ijerph-320180) (section.2)]
>>
endobj
18 0 obj
<<
/Kids [71 0 R 72 0 R 73 0 R 74 0 R 75 0 R]
/Limits [(section.3) (table.3)]
>>
endobj
19 0 obj
<<
/A <<
/S /URI
/Type /Action
/URI (http://www.mdpi.com/journal/ijerph)
>>
/Border [0 0 0]
/C [0 1 1]
/H /I
/Rect [75.539 757.64 221.788 793.648]
/Subtype /Link
/Type /Annot
>>
endobj
20 0 obj
<<
/A <<
/S /URI
/Type /Action
/URI (http://www.mdpi.com)
>>
/Border [0 0 0]
/C [0 1 1]
/H /I
/Rect [474.736 757.64 519.737 793.648]
/Subtype /Link
/Type /Annot
>>
endobj
21 0 obj
<<
/A <<
/S /URI
/Type /Action
/URI (https://orcid.org/0000-0002-2962-9254)
>>
/Border [0 0 0]
/C [0 1 1]
/H /I
/Rect [215.274 659.466 228.102 674.364]
/Subtype /Link
/Type /Annot
>>
endobj
22 0 obj
<<
/A <<
/S /URI
/Type /Action
/URI (https://orcid.org/0000-0002-5805-834X)
>>
/Border [0 0 0]
/C [0 1 1]
/H /I
/Rect [374.731 659.466 387.559 674.364]
/Subtype /Link
/Type /Annot
>>
endobj
23 0 obj
<<
/A <<
/S /URI
/Type /Action
/URI (http://www.mdpi.com/1660-4601/15/7/1450?type=check_update&version=1)
>>
/Border [0 0 0]
/C [0 1 1]
/H /I
/Rect [467.499 528.758 519.737 550.287]
/Subtype /Link
/Type /Annot
>>
endobj
24 0 obj
<<
/A <<
/D (cite.B1-ijerph-320180)
/S /GoTo
>>
/Border [0 0 0]
/C [0 1 0]
/H /I
/Rect [508.582 182.085 515.556 191.012]
/Subtype /Link
/Type /Annot
>>
endobj
25 0 obj
<<
/A <<
/S /URI
/Type /Action
/URI (http://dx.doi.org/10.3390/ijerph15071450)
>>
/Border [0 0 0]
/C [0 1 1]
/H /I
/Rect [254.937 38.684 340.121 50.141]
/Subtype /Link
/Type /Annot
>>
endobj
26 0 obj
<<
/A <<
/S /URI
/Type /Action
/URI (http://www.mdpi.com/journal/ijerph)
>>
/Border [0 0 0]
/C [0 1 1]
/H /I
/Rect [405.255 38.684 519.737 50.141]
/Subtype /Link
/Type /Annot
>>
endobj
27 0 obj
<<
/Length 681
/Filter /FlateDecode
>>
stream
HUMo0WҮY
h ;;ha~e9d@7(=T
@aѿ
&P"GÌVСk9R``4QBQhd3s$fa͆U-l#ZC*;H*|J(bM&k~pq컶Jt)]tǀqUon[GۮZ8,vQRPb˶UmrnY(BqN
$${|C
پ@5D'^ٕWcqAk
ڷ e]ΏP}Y?S<#rDg}^{>1 {80bUSq@0W
5B/
$21"oț#ޚk
Ɏ.^.=y,`ZKgzLN;dzהLYo(εcҿ#QX( `&$qBL\SV~$NJ
A1ݑeOmſ
endstream
endobj
28 0 obj
<<
/Length 710
/Filter /FlateDecode
>>
stream
HUɎ0+xZվ\3
CLStCF8ߗ,xREEeA9'Tpȴ9u-
qkXsCU XoUqE}^aeXk1Z ޖH&#)Ȧ/:ix;yΈj 2a8fnͺH`z|yqO87xG+Xjp΅-g#$((*DZZo ɦ [z^טUF,aLHgGjȠċZq\u
180bW:(7n aD$ZAԊJmBn3r[5Z
\տbN^LC`VAx=vM7f)8Zsʽ옏 68CD9Tc7RtG9?SH_Gx[Km:mZXMmkԢ>5A}{yjN