Back to search

Georgia Institute of Technology

Toward The Democratization of Industrial Machine Learning

Abstract

dc:description.abstract

Machine learning (ML) is an incredibly powerful tool for learning representations from data and, given the ever-growing amount of data made available by the Industrial Internet of Things (IIoT), machine learning is of increasing interest in manufacturing. The confluence of ML and IIoT has ushered in the 4th Industrial Revolution and has touched nearly every aspect of manufacturing operations, from machine health to part quality to supply chains. Despite its already proven success, deploying an ML model in a new application requires overcoming several barriers associated with resource allocation, data acquisition, and access to ML expertise. To lower these barriers to entry, there has been significant work in the machine learning field, including scalable computing platforms, open-source deep-learning software frameworks, and publicly available fundamental models. While these advances can also be used by ML practitioners in industrial or manufacturing areas, this field has several unique challenges that have limited widespread adoption of ML. These challenges include feature selection from unique data sources, model selection within a new domain, and resource constrained model deployment. This dissertation advances the state-of-the-art of democratized machine learning in industrial applications through three studies. First, to make full use of the advances from image classification, advanced convolution neural network architectures, developed for image classification, were adapted for sensor data to estimate remaining useful life (RUL) of turbofan engines. All architectures tested were found to perform well with model scores between 5.8 and 6.2 cycles and no statistically significant differences between them. What was statistically significant was the differences between model inputs that yielded differences in model scores on the order of 1.5 cycles. Additionally, it was observed that the model accuracy improved as the true RUL decreased, indicating that prognostics efforts will be more effective if they focus on engines later in life. Partially motivated by the impact that the model inputs had in the first study, in the next study, a methodology to use Bayesian optimization and image similarity was introduced for selecting the optimal time-frequency representation of sensor data without previous domain knowledge. This methodology was tested on two datasets whose objectives were to classify bearing health and classify the standoff distance in wire arc additive manufacturing. When the produced time-frequency representations were used as model inputs to simple convolutional neural networks, they yielded accuracies of 100% and 94%, indicating the appropriateness of the TFRs selected. Finally, an architecture was proposed and tested to use anomaly detection as an input filter on edge devices to determine when data should be passed to cloud deployed supervised models. Building off the preprocessing from the previous studies, which already reduces the data size by 94% and 75% for the turbofan and bearing datasets, the anomaly detector further reduces the volume of data by 73%. Additionally, for the turbofan dataset, because the anomaly detector primarily identifies samples later in an engine’s life, the performance of the supervised model improves from 5.8 cycles to 1.5 cycles. Together, the technical work in this dissertation develops an understanding of how to integrate ML with industrial applications, and this knowledge can be leveraged by manufacturers of all sizes to improve manufacturing processes and produced goods.

Degree

thesis:*
Level thesis:degree_level
Doctoral
Department dc:contributor.department
Mechanical Engineering
Grantor dc:publisher
Georgia Institute of Technology
Year dc:date.issued
2024

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • DeVol, Nathaniel
Advisors dc:contributor.advisor
  • Saldaña, Christopher J.
  • Fu, Katherine
Committee members dc:contributor.committeemember
  • Kurfess, Thomas
  • Henderson, Andrew
  • Ahuett-Garza, Horacio

Subjects

dc:subject × 5

Rights

Language dc:language.iso
en_US

Identifiers

dc:identifier.*
Handle dc:identifier.uri
https://hdl.handle.net/1853/77722

Chain of custody

source
Harvested from
Georgia Tech
Base URL
repository.gatech.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

DeVol, Nathaniel. Toward The Democratization of Industrial Machine Learning. Doctoral thesis, Georgia Institute of Technology, 2024. https://hdl.handle.net/1853/77722