東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Source Apportionment of Combustion G...

Rutherford, Jay W.

FindBook

Google Book

Amazon

博客來

Source Apportionment of Combustion Generated Particulate Matter Air Pollution Using Excitation Emission Matrix Fluorescence Spectroscopy and Machine Learning.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Source Apportionment of Combustion Generated Particulate Matter Air Pollution Using Excitation Emission Matrix Fluorescence Spectroscopy and Machine Learning./
作者:	Rutherford, Jay W.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2019,
面頁冊數:	116 p.
附註:	Source: Dissertations Abstracts International, Volume: 81-09, Section: B.
Contained By:	Dissertations Abstracts International81-09B.
標題:	Environmental health. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=27669583
ISBN:	9781392564905

Source Apportionment of Combustion Generated Particulate Matter Air Pollution Using Excitation Emission Matrix Fluorescence Spectroscopy and Machine Learning.
Rutherford, Jay W.

Source Apportionment of Combustion Generated Particulate Matter Air Pollution Using Excitation Emission Matrix Fluorescence Spectroscopy and Machine Learning. - Ann Arbor : ProQuest Dissertations & Theses, 2019 - 116 p.

Source: Dissertations Abstracts International, Volume: 81-09, Section: B.

Thesis (Ph.D.)--University of Washington, 2019.

This item must not be sold to any third party vendors.

Exposure to particulate matter (PM) air pollution is the world's largest environmental health risk accounting for millions of premature deaths and disability-adjusted life years annually. PM originates from natural and anthropogenic sources such as dust from soil, combustion engines, and forest fires, among many others. PM exposure is quantified by measuring its mass concentration in air. This measurement alone does not identify the sources of PM exposure, which can inform effective mitigation strategies and allow for studying source-specific health effects. There are several options for source apportionment (e.g. GC-MS and X-ray fluorescence), but they are costly and time consuming to conduct. Alternative methods for source apportionment using low-cost techniques would be beneficial to the study of air pollution and its health effects. In this dissertation, I develop a method for source apportionment of combustion generated PM using fluorescent Excitation Emission Matrix (EEM) fluorescent spectroscopy and machine learning.First, I collected PM samples from combustion sources of concern to human health in the laboratory. I analyzed cyclohexane extracts of cigarette smoke, diesel exhaust and wood smoke by EEM fluorescent spectroscopy and using the World Health Organization's guideline for annual mean PM exposure of 10 µg/m3 as a basis of comparison I show EEM is sensitive enough to detect combustion generated PM at levels well below those of concern to human health.Next, mixtures of the same laboratory sources are analyzed using EEM. Combining measurements of the individual sources with those of mixtures, I apply several machine learning techniques and a simple linear model to perform source apportionment and identification from the mixtures and compare the results. A convolutional neural network (CNN) is found to have the best performance of all methods investigated. I describe in detail the architecture and data augmentation approach used for the CNN.Finally, the EEM-Machine Learning approach is used for source apportionment of environmental samples. Results and filter samples from an exposure assessment panel study are used for this analysis. The samples were analyzed using X-ray fluorescence and source apportionment was conducted using Positive Matrix Factorization. Filters, archived in a freezer, were extracted with cyclohexane and analyzed by EEM. The resulting EEM spectra and source contribution estimates from PMF were used as training data for the application of machine learning. A CNN with the same architecture as applied to the laboratory samples and Principal Component Regression showed similar results in predicting contributions from combustion generated PM. These methods were able to reproduce the XRF-PMF results with R2 values as high as 0.84 for vegetative burning and 0.52 for traffic emissions.

ISBN: 9781392564905Subjects--Topical Terms:

543032
Environmental health.
Subjects--Index Terms:

EEM fluorescence

Source Apportionment of Combustion Generated Particulate Matter Air Pollution Using Excitation Emission Matrix Fluorescence Spectroscopy and Machine Learning.
LDR:04147nmm a2200361 4500 001 2272992
005 20201105110317.5
008 220629s2019 ||||||||||||||||| ||eng d
020 $a 9781392564905
035 $a (MiAaPQ)AAI27669583
035 $a AAI27669583
040 $a MiAaPQ $c MiAaPQ
100 1 $a Rutherford, Jay W. $3 3550416
245 1 0 $a Source Apportionment of Combustion Generated Particulate Matter Air Pollution Using Excitation Emission Matrix Fluorescence Spectroscopy and Machine Learning.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2019
300 $a 116 p.
500 $a Source: Dissertations Abstracts International, Volume: 81-09, Section: B.
500 $a Advisor: Novoselov, Igor V.;Posner, Jonathan D.
502 $a Thesis (Ph.D.)--University of Washington, 2019.
506 $a This item must not be sold to any third party vendors.
506 $a This item must not be added to any third party search indexes.
520 $a Exposure to particulate matter (PM) air pollution is the world's largest environmental health risk accounting for millions of premature deaths and disability-adjusted life years annually. PM originates from natural and anthropogenic sources such as dust from soil, combustion engines, and forest fires, among many others. PM exposure is quantified by measuring its mass concentration in air. This measurement alone does not identify the sources of PM exposure, which can inform effective mitigation strategies and allow for studying source-specific health effects. There are several options for source apportionment (e.g. GC-MS and X-ray fluorescence), but they are costly and time consuming to conduct. Alternative methods for source apportionment using low-cost techniques would be beneficial to the study of air pollution and its health effects. In this dissertation, I develop a method for source apportionment of combustion generated PM using fluorescent Excitation Emission Matrix (EEM) fluorescent spectroscopy and machine learning.First, I collected PM samples from combustion sources of concern to human health in the laboratory. I analyzed cyclohexane extracts of cigarette smoke, diesel exhaust and wood smoke by EEM fluorescent spectroscopy and using the World Health Organization's guideline for annual mean PM exposure of 10 µg/m3 as a basis of comparison I show EEM is sensitive enough to detect combustion generated PM at levels well below those of concern to human health.Next, mixtures of the same laboratory sources are analyzed using EEM. Combining measurements of the individual sources with those of mixtures, I apply several machine learning techniques and a simple linear model to perform source apportionment and identification from the mixtures and compare the results. A convolutional neural network (CNN) is found to have the best performance of all methods investigated. I describe in detail the architecture and data augmentation approach used for the CNN.Finally, the EEM-Machine Learning approach is used for source apportionment of environmental samples. Results and filter samples from an exposure assessment panel study are used for this analysis. The samples were analyzed using X-ray fluorescence and source apportionment was conducted using Positive Matrix Factorization. Filters, archived in a freezer, were extracted with cyclohexane and analyzed by EEM. The resulting EEM spectra and source contribution estimates from PMF were used as training data for the application of machine learning. A CNN with the same architecture as applied to the laboratory samples and Principal Component Regression showed similar results in predicting contributions from combustion generated PM. These methods were able to reproduce the XRF-PMF results with R2 values as high as 0.84 for vegetative burning and 0.52 for traffic emissions.
590 $a School code: 0250.
650 4 $a Environmental health. $3 543032
650 4 $a Chemical engineering. $3 560457
650 4 $a Analytical chemistry. $3 3168300
653 $a EEM fluorescence
653 $a Machine learning
653 $a Source apportionment
690 $a 0470
690 $a 0542
690 $a 0486
710 2 $a University of Washington. $b Chemical Engineering. $3 3178392
773 0 $t Dissertations Abstracts International $g 81-09B.
790 $a 0250
791 $a Ph.D.
792 $a 2019
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=27669583