Application of Bayesian Machine Learning and Uncertainty Quantification in Structured Light System and Pipeline External Corrosion Assessment
Date of Award
12-1-2023
Degree Name
Ph.D. in Electrical Engineering
Department
Department of Electrical and Computer Engineering
Advisor/Chair
Advisor: Hui Wang
Abstract
Bayesian methods offer a powerful framework for predicting and quantifying uncertainty in various practical systems. This study explores the application of Bayesian approaches in two distinct domains: structured light (SL) systems and pipeline corrosion analysis. In structured light systems, Bayesian method is utilized to build a complete coordinate uncertainty model from the root cause. Details on sources of aleatoric and epistemic uncertainty and its propagation to final reconstructed coordinates is discussed. Certain practical implementation issues and possible solutions are also discussed. Conversely, in pipeline corrosion analysis, Bayesian methodologies are employed to model uncertainty derived from acquired data. This application involves leveraging Bayesian inference to understand, model, and mitigate uncertainties arising from environmental factors and material behaviors. Through these two applications, Bayesian methods prove instrumental in navigating and managing uncertainties in practical systems, offering insights, predictions, and solutions vital for decision-making and system optimization.
Keywords
structured light, bayesian inference, uncertainty quantification, pipeline corrosion, pipeline reliability, aleatoric uncertainty, epistemic uncertainty
Rights Statement
Copyright © 2023, author.
Recommended Citation
Sreelakshmi, Sreeharan, "Application of Bayesian Machine Learning and Uncertainty Quantification in Structured Light System and Pipeline External Corrosion Assessment" (2023). Graduate Theses and Dissertations. 7362.
https://ecommons.udayton.edu/graduate_theses/7362
