Theses and Dissertations

Advisor

Tian, Wenmeng

Committee Member

Bian, Linkan

Committee Member

Ma, Junfeng

Committee Member

Priddy, Matthew W.

Date of Degree

8-13-2024

Original embargo terms

Embargo 6 months

Document Type

Dissertation - Open Access

Major

Industrial & Systems Engineering

Degree Name

Doctor of Philosophy (Ph.D.)

College

James Worth Bagley College of Engineering

Department

Department of Industrial and Systems Engineering

Abstract

Metal-based additive manufacturing (AM) has emerged as a cutting-edge technology for fabricating complex geometries with high precision. However, the major challenges to the wider adoption of metal AM technologies are process uncertainty-induced quality issues. Consequently, there is an urgent need for fast and reliable certification techniques for AM components, which can be achieved by leveraging Artificial Intelligence (AI)-enabled modeling. Developing a robust AI-enabled model presents a significant challenge because of the costly and time-intensive nature of acquiring diverse and high volume of datasets. In this context, the data-sharing attributes of Manufacturing-as-a-Service (MaaS) platforms can facilitate the development of AI-enabled certification techniques in a collaborative manner. However, sharing process data poses critical concerns about protecting users’ intellectual property and privacy since it contains confidential product design information. To address these challenges, the overarching goal of this research is to investigate how process data and process physics can be leveraged to develop in-situ component certification techniques focusing on data privacy for metal AM systems. This dissertation aims to address the need for novel quality monitoring methodologies by utilizing diverse data sources derived from a range of printed samples. Specifically, the research effort focuses on 1) the use of in-situ thermal history data and ex-situ X-ray computed tomography data for real-time layer-wise anomaly detection method development by analyzing the morphological dynamics of melt pool images; 2) the development of a framework to evaluate the design information disclosure of various thermal history-based feature extraction methods for anomaly detection; and 3) the privacy-preserving and utility-aware adaptive AM data deidentification method development that takes thermal history data as input.

Available for download on Sunday, December 15, 2024

Share

COinS