Date of Award
5-2023
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Civil Engineering
Committee Chair/Advisor
Dr. Mashrur Chowdhury
Committee Member
Dr. Sakib Khan
Committee Member
Dr. Bryan Riley
Abstract
The environmental impacts of global warming driven by fugitive methane (CH4) emissions have catalyzed significant research initiatives in developing novel technologies that enable proactive and rapid detection of CH4 emissions. This study evaluated the performance of data-driven machine learning (ML) models using support vector machines (SVM) to detect the presence of trace CH4 emissions and the corresponding intensity amongst various meteorological conditions. The author used simulation data comprising various meteorological parameters such as temperature, relative humidity, wind speed, water vapor, pressure, precipitation rate, and a parameter possessing trace concentrations of CH4 emissions. The novelty of the SVM models developed in this work lies in the ability to (i) detect the presence of trace concentrations of CH4 as a classification task, and (ii) predict the corresponding intensity of CH4 levels using AI-based regression analysis. The metrics used to assess the classification performance for SVM CH4 detection were accuracy, F1-score, and area under the receiver operating characteristic curve (AUC-ROC) with results for the best-performing model being 0.89, 0.89, and 0.94, respectively. On the other hand, the root mean squared error (RMSE) and mean absolute percentage error (MAPE) scores were used to evaluate the regression model performance for trace CH4 intensity prediction, achieving scores of 0.98 and 1.22 respectively, thereby demonstrating the reliable low error probability of the regression model in forecasting levels of trace CH4 emissions.
Recommended Citation
Pollard, Jacquan, "Development and Evaluation of Machine Learning Models for Fugitive Methane Detection and Intensity Prediction" (2023). All Theses. 4065.
https://open.clemson.edu/all_theses/4065