Doctor of Philosophy (PhD)



Document Type



This dissertation investigates cutting-edge AI-based approaches to enhance Nondestructive Evaluation in Additive Manufacturing (AM). As a larger workflow that aims to design complex alloys and shape memory polymers for AM, this effort tries to leverage machine learning to achieve performance and structural integrity goals in the material characterization stage. The dissertation comprises three key test cases:

1. X-ray CT Data Analysis Using Machine Learning

This section identifies the limitations of manual pixel-level segmentation in X-ray CT data analysis and introduces zero-shot learning. A self-supervised approach is proposed, eliminating manual segmentation and enhancing crack detection and prediction.

2. Neutron Imaging of MELD Samples

Utilizing neutron imaging, this section evaluates large Al 6061 samples produced through Additive Friction Stir Deposition (MELD). It reveals layering structures and identifies hydrogen contamination sources, shedding light on its impact on mechanical properties and grain structure using different modalities of neutron imaging.

3. Combined X-ray/Neutron Imaging

A novel combined X-ray/neutron imaging instrument is introduced, offering unique AM quality assessment capabilities. The instrument's feasibility is justified, and a data fusion pipeline is developed for clustering phases within Wire Arc Additive Manufacturing (WAAM) samples.

In summary, this dissertation advances AM NDE through innovative imaging and machine learning techniques. It addresses current challenges and opens new avenues for enhancing AM component reliability and structural integrity.



Committee Chair

Guo, Shengmin

Available for download on Sunday, November 01, 2026