Towards automated and explorative characterization of nano-energetic material response to directed energy
Authors
Meeting name
Sponsors
Date
Journal Title
Format
Thesis
Subject
Abstract
Nanoenergetic materials offer high-density energy storage that may react to produce heat and release gaseous products. In order to utilize aluminum-based nanoenergetic materials effectively, the mechanisms by which aluminum fuel reacts to escape a passivating aluminum shell must be better understood. These mechanisms are affected by numerous parameters in both the laser-based photothermal heating setup to incite reactions as well as in the shape, size, and composition of the material itself. The vast number of parameters greatly increases the size of the search space when seeking a quantitative mapping between reaction parameters and reaction type. This thesis aims to accelerate this search through the creation of an autonomous image pro- cessing pipeline for nanoenergetic material reaction imagery, as well as a subsequent reaction classifier. The imagery is captured at 100x magnification utilizing optical microscopy and, various preprocessing methods are explored to aid in highlighting distinctive image features for characterizing reaction type. To perform the classification, we began with hand-crafted features and ensemble machine learning. Building on this, we modified a state-of-the-art transformer-based model for change detection to perform reaction classification. In doing so, we validated an abundant analog for nanoenergetic material, cell nuclei, to utilize for pre-training. Finally, we investigate texture-based image features of nanoenergetic material in scanning electron microscope (SEM) imagery. By leveraging this texture analysis, we can effectively isolate nanoenergetic material clusters within SEM imagery enabling autonomous laser targeting and imaging in an SEM environment.
Table of Contents
PubMed ID
Degree
M.S.