Date of Graduation

8-2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Engineering (PhD)

Degree Level

Graduate

Department

Industrial Engineering

Advisor/Mentor

Chase Rainwater

Committee Member

Ed Pohl

Second Committee Member

Xiao Liu

Third Committee Member

Jackson Cothren

Keywords

absolute visual localization, deep learning, geolocalization, geolocation, repairable systems

Abstract

Deep learning - the use of large neural networks to perform machine learning - has transformed the world. As the capabilities of deep models continue to grow, deep learning is becoming an increasingly valuable and practical tool for industrial engineering. With its wide applicability, deep learning can be turned to many industrial engineering tasks, including optimization, heuristic search, and functional approximation. In this dissertation, the major concepts and paradigms of deep learning are reviewed, and three industrial engineering projects applying these methods are described. The first applies a deep convolutional network to the task of absolute aerial geolocalization - the regression of real geographic coordinates from aerial photos - showing promising results. Next, continuing on this work, the features and characteristics of the deep aerial geolocalization model are further studied, with implications for future applications and methodological improvements. Lastly, a deep learning model is developed and applied to a difficult rare event problem of predicting failure times in oil and natural gas wells from process and site data. Practical details of applying deep learning to this sort of data are discussed, and methodological principles are proposed.

Share

COinS