Abstract
dc:description.abstract3D printers, specifically FDM extrusion-type printers, emit various types of acoustic signals. These acoustic signals are emitted from the stepper motors that control the printhead (nozzle) movement in the X, Y, and Z axes. Due to the method of current delivery to these motors, the acoustics emitted have characteristic frequency spectra and are easily identified. Of course, depending on the type of object being printed, the time and frequency domain content are drastically changed. As a result, this allows for insight into, and possibly a deduction of the object being printed. If the audio data is processed properly, and the correct classification or clustering algorithms are used, it is possible to ascertain the object of print with no prior knowledge beyond the acoustic signals emitted. The main goals of this study consist of: demonstrating that categorization of basic print shapes using 3D printer acoustics is possible, comparing preprocessing methods’ accuracies and runtimes for both K-means clustering and random forest machine learning algorithms as functions of recording distance, and recommending the most effective preprocessing methods relative to accuracy and runtime. With this technology, nonproliferation efforts could be significantly advanced, and additionally this may be useful for developing further methods for other types of 3D printer signal analysis.
Degree
thesis:*- Level thesis:degree_level
- Masters
- Department dc:contributor.department
- Mechanical Engineering
- Grantor dc:publisher
- Georgia Institute of Technology
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Nunnally, Hugh P.
- Advisor dc:contributor.advisor
-
- Biegalski, Steven R.
- Committee members dc:contributor.committeemember
-
- Erickson, Anna
- Haas, Derek
Subjects
dc:subject × 1Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/1853/77833