Thesis
Application of pattern recognition techniques to aid in detecting degradation of a pulsed power system
Master of Science (M.S.), Drexel University
Jun 2018
DOI:
https://doi.org/10.17918/D81T2R
Abstract
Pulsed power is the release of energy over a small interval of time, resulting in a large peak in power at the load. MERCURY is a 2.2 TW pulsed power system used in experiments at the Naval Research Lab. During high power experiments, it was noted that the coating on the inductive cells was damaged. Throughout exploration of the system, researchers determined that when the trigger misfires, the switches can be damaged, which causes ringing and energy dissipation throughout the entire system. It was desired to use the data from the high power experiments to create a tool to automatically examine all of the signals associated with one shot or machine cycle of the pulse power system. A data structure was designed to organize and analyze the data associated with one shot. With this structure in place, it is easy to extract features about the system and explore degradation trends. By exploring these trends, an algorithm was designed to identify and predict when the pulsed power system begins to experience degradation. In this way, damage to the equipment could be prevented. Pattern recognition techniques such as nonlinear unsupervised clustering were applied to data from the MERCURY system, in order to predict and diagnose degradation of equipment.
Metrics
42 File views/ downloads
32 Record Views
Details
- Title
- Application of pattern recognition techniques to aid in detecting degradation of a pulsed power system
- Creators
- Sarah Peachey - DU
- Contributors
- Christopher W. Peters (Advisor) - Drexel University (1970-)Thomas A. Chmielewski (Advisor) - Drexel University (1970-)
- Awarding Institution
- Drexel University
- Degree Awarded
- Master of Science (M.S.)
- Publisher
- Drexel University; Philadelphia, Pennsylvania
- Number of pages
- xi, 63 pages
- Resource Type
- Thesis
- Language
- English
- Academic Unit
- College of Engineering (1970-2026); Electrical (and Computer) Engineering (1970-2026); Drexel University
- Other Identifier
- 8186; 991014632191004721